2025-08-19to30科研追新

2025-08-18 19:37:26 Monday ~ 2025-08-30 12:23:19 Saturday

1. 源数据

1.1 公众号

1.1.1 量子位

1. 港股AGI第一股“云知声”首战告捷：大模型贡献1亿收入，单客价直线提升116.2%，AI保险业务暴涨1386.8%
2. 不愧是中国机器人，乒乓打得太6了
3. 吴恩达最新来信：是时候关注并行智能体了
4. 10年前押中英伟达：这位复旦学霸如何用AI Agent重新定义投资
5. 老黄又投了一个核电站
6. Nano banana手办玩法火爆出圈！无需抽卡，效果惊了(°o°)
7. 蚂蚁专用模型超越o3！仅用2K训练样本刷新医疗AI榜单纪录
8. 马斯克入局AI编程！xAI新模型限时免费用：256K上下文，主打一个速度快
9. 腾讯混元最新开源：一键生成电影级音效，性能表现全面SOTA
10. 小米新系统和iPhone联动了
11. 一帮人All in AI，让搞体育的先赚到钱了
12. AI搜索MCP服务来了，Agent直接链接实时信息！刚刚，百度智能云打出了张“王牌”
13. ChatGPT后遗症来了！人类日常聊天越来越AI化
14. 啊？猫猫也会老年痴呆
15. AI人才争夺战加大薪资差距，OpenAI前副总裁：能留住人才是最重要的
16. 对话逐际动力张巍：造机器人很容易，关键是用起来
17. 波士顿动力机器狗侧空翻炸场！穿轮滑鞋照样能翻
18. OpenAI和Anthropic罕见互评模型：Claude幻觉明显要低
19. 陈丹琦有了个公司邮箱，北大翁荔同款
20. 老黄太难了！英伟达Q2营收467亿美元创纪录，股价盘后还跌了5%
21. 小扎高薪挖来的人又跳回OpenAI了！首席科学家赵晟佳也要回去
22. 北大南开数学家解决著名“十杯马天尼”问题：更统一、更优雅的证明
23. 空间智能卡脖子难题被杭州攻克！难倒GPT-5后，六小龙企业出手了
24. 谷歌认领最强AI版Photoshop！现在人人可用，效果确实强悍
25. Claude for Chrome来了！可作为浏览器扩展程序直接使用
26. 新iPhone的AI怎么样，得看苹果最近的收购了
27. 破解人机协作密码：工作技能拆成两层，AI执行人类决策成功率狂飙 | ICML 2025
28. 数字技术工人已到岗！时序大模型+Agent已掌握了工厂生产管控技术，比人类更懂工况
29. DeepSeek“极你太美”bug，官方回应了
30. 阿里开源14B电影级视频模型！实测来了：免费可玩，单次生成时长可达分钟级
31. 英伟达韩松团队新作：具有后神经架构搜索的高效语言模型
32. GPT-5通关《宝可梦水晶》创纪录！9517步击败赤爷，效率碾压o3三倍！
33. 视觉Token注入CLIP语义，走向多模态理解与生成新范式
34. 最新智能体自动操作手机电脑，10个榜单开源SOTA全拿下｜通义实验室
35. 最高提效8倍！腾讯游戏发布专业游戏AI大模型，美术师做动画不用辣么“肝”了
36. Karpathy氛围编程最新指南！三层AI编程结构：顺境Cursor，逆境Claude，绝境GPT-5 Pro
37. AI视频生成新品实测：这怎么不算影院级呢？
38. 为防AI刷题，Nature等顶刊最新封面被做成数据集，考验模型科学推理能力|上海交通大学
39. 首个接入GPT-5的视频Agent！一句话生成商业级广告大片，分镜配音字幕等全包了
40. GPT-5系统提示词被曝，足足15000 tokens！
41. 诺贝尔物理学成果48年后终获数学证明！中科大少年班尹骏又出现了
42. 和图灵机相关的这个数字，已经大到整个宇宙原子都容不下了
43. 告别“炼丹玄学”：上海AI实验室推出首个大模型数据竞技场OpenDataArena
44. 刚刚，马斯克开源Grok 2.5：中国公司才是xAI最大对手
45. 让AI作画自己纠错！随机丢模块就能提升生成质量，告别塑料感废片
46. 阿里全新AI IDE现在免费用：超强上下文理解，覆盖整个代码库
47. 首个故事可视化综合评估框架来了！80个故事单元53种类别，20种技术方案全面对比
48. GPT-5 Pro独立做数学研究！读论文后给出更精确边界，OpenAI总裁：这是生命迹象
49. 开源复现o3图像思考！快手让AI不再被动看图，模型自主生成代码调用工具
50. 字节突然开源Seed-OSS，512K上下文碾压主流4倍长度！推理能力刷新纪录
51. 突破Agent长程推理效率瓶颈！MIT&新加坡国立联合推出强化学习新训练方法
52. 实测DeepSeek V3.1，不止拓展上下文长度
53. 思维链可无限延伸了，MIT等打破大模型上下文天花板
54. 英伟达开源9B参数小模型，比Qwen3快6倍
55. 突破Claude-4编程上限！自进化Agent框架拿下新SOTA，底模越好性能越高，已开源

1.1.2 机器之心

1. 23岁小哥被OpenAI开除，成立对冲基金收益爆表，165页论文传遍硅谷
2. 在美国，打工人越老越吃香，22-25岁新人最先被AI淘汰
3. 你能永远陪我聊天吗？复旦&微软提出StableAvatar: 首个端到端无限时长音频驱动的人类视频生成新框架!
4. 合成数据的「毒」与「药」，模型崩溃有何新解？
5. 清华崔鹏团队开源LimiX：首个结构化数据通用大模型，性能超越SOTA专用模型
6. 谢赛宁回忆七年前OpenAI面试：白板编程、五小时会议，面完天都黑了
7. AI Agent组团搞事：在你常刷的App里，舆论操纵、电商欺诈正悄然上演
8. Grok代码模型来了：限时免费用，速度超级快
9. 杜克大学、Zoom推出LiveMCP‑101：GPT‑5表现最佳但未破60%，闭源模型Token效率对数规律引关注
10. DeepSeek刚提到FP8，英伟达就把FP4精度推向预训练，更快、更便宜
11. We-Math 2.0：全新多模态数学推理数据集 × 首个综合数学知识体系
12. 打破瓶颈，让RAG学会思考：中科大、智源等发布推理检索框架BGE-Reasoner
13. Agentic Deep Research新范式，推理能力再突破，可信度增加，蚂蚁安全团队出品
14. 打磨7年，李航新书《机器学习方法（第2版）》发布，有了强化学习，赠书20本
15. 「开发者私下更喜欢用GPT-5写代码」，Claude还坐得稳编程王座吗？
16. ChatGPT到底学了多少「污言秽语」？清华团队首提大语言模型中文语料污染治理技术
17. 唯快不破：上海AI Lab 82页综述带你感受LLM高效架构的魅力
18. 仅靠5000+样本，全新强化学习范式让30B轻松击败671B的DeepSeek V3
19. Chain-of-Agents: OPPO推出通用智能体模型新范式，多榜单SOTA，模型代码数据全开源
20. 全球首款AI原生游戏引擎再进化：GTA6再不来，我们就AI一个
21. KDD 2025 Best Paper Runner-Up | EI-BERT：超紧凑语言模型压缩框架
22. 从繁杂技巧到极简方案：ROLL团队带来RL4LLM新实践
23. ICCV 2025 | 打造通用工具智能体的基石：北大提出ToolVQA数据集，引领多模态多步推理VQA新范式
24. ICCV 2025 | ECD：高质量合成图表数据集，提升开源MLLM图表理解能力
25. 击败Meta登榜首：推理增强的文档排序模型ReasonRank来了
26. DiT在数学和形式上是错的？谢赛宁回应：不要在脑子里做科学
27. dLLM的「Free Lunch」！浙大&蚂蚁利用中间结果显著提升扩散语言模型
28. 强化学习之父Richard Sutton最新演讲揭示OaK架构：通向超级智能的八步愿景

1.1.3 新智元

1. 国产黑马自优化「超级大脑」，全闭环Agent杀疯！一站式AI原生基建来了
2. 从需求分析到代码生成，LLM都能干点啥？一文读懂291个软工Benchmark！
3. 2025临界点：AI智商超越人类，经济规则即将改写
4. GPT-5系统提示词突遭泄露，17803 token曝光OpenAI小心思！
5. OpenAI用GPT-4b攻克诺奖难题！人体细胞「返老还童」，逆转效率飙升50倍
6. 比GPT-5还准？AIME25飙到99.9%刷屏，开源模型首次！
7. 刚刚，大模型棋王诞生！40轮血战，OpenAI o3豪夺第一，人类大师地位不保？
8. GPT-5点赞！八大顶尖机构发布「自进化智能体」全面综述

1.1.4 AGI Hunt

1. OpenAI与Anthropic罕见合作：竞争对手联手测试AI安全
2. Gemini 3 本周发布？
3. Hugging Face 推出九大 AI 课程，免费、全面【收藏】
4. 字节发布全球首个预测未来基准FutureX，Grok-4 拿下冠军
5. DeepSeek 官宣V3.1：迈向 Agent 时代的第一步！

1.1.5 其他

1.2 Arxiv

1.2.1 Computation and Language

From：https:// /arxiv/cs.CL

From：https://arxiv.org/list/cs.CL/recent

• [1] arXiv:2508.21051 [pdf, html, other]

  Enabling Equitable Access to Trustworthy Financial Reasoning 促进对可信金融推理的公平获取William Jurayj, Nils Holzenberger, Benjamin Van Durme William Jurayj、Nils Holzenberger、Benjamin Van DurmeSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；计算机与社会 (cs.CY)

• [2] arXiv:2508.21049 [pdf, html, other]

  Re-Representation in Sentential Relation Extraction with Sequence Routing Algorithm 在句子关系抽取中使用序列路由算法的重新表征Ramazan Ali Bahrami, Ramin Yahyapour Ramazan Ali Bahrami，Ramin YahyapourComments: Presented in 8th International Conference on Natural Language and Speech Processing (ICNLSP), 25-27 August 2025, SDU, Odense, Denmark 注释：在第八届国际自然语言与语音处理会议（ICNLSP），2025 年 8 月 25-27 日，丹麦奥登塞南丹大学（SDU）上宣读Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [3] arXiv:2508.21024 [pdf, other]

  An Agile Method for Implementing Retrieval Augmented Generation Tools in Industrial SMEs 一种在工业中小企业实现检索增强生成工具的敏捷方法Mathieu Bourdin, Anas Neumann, Thomas Paviot, Robert Pellerin, Samir LamouriComments: 20 pages, 3 figures 注释：20 页，3 幅图Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；信息检索 (cs.IR)

• [4] arXiv:2508.21004 [pdf, other]

  Lethe: Purifying Backdoored Large Language Models with Knowledge Dilution Lethe：通过知识稀释净化带后门的大型语言模型Chen Chen, Yuchen Sun, Jiaxin Gao, Xueluan Gong, Qian Wang, Ziyao Wang, Yongsen Zheng, Kwok-Yan Lam 陈晨，孙昱辰，高嘉鑫，龚雪卵，王倩，王子尧，郑永森，林郭仁言Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [5] arXiv:2508.20973 [pdf, html, other]

  ProactiveEval: A Unified Evaluation Framework for Proactive Dialogue Agents ProactiveEval：一个用于主动对话代理的统一评估框架Tianjian Liu, Fanqi Wan, Jiajian Guo, Xiaojun Quan 刘天舰，万凡琦，郭佳鉴，权晓军Comments: 21 pages, 6 Figures 注释：21 页，6 幅图Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；人机交互 (cs.HC)

• [6] arXiv:2508.20944 [pdf, html, other] [6] arXiv:2508.20944 [ pdf，html，other]

  STARE at the Structure: Steering ICL Exemplar Selection with Structural Alignment 注视结构：通过结构对齐引导 ICL 示例选择Jiaqian Li, Qisheng Hu, Jing Li, Wenya Wang 李佳倩，胡启胜，李静，王文雅Comments: EMNLP 2025 Main 评论：EMNLP 2025 主会议Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [7] arXiv:2508.20931 [pdf, html, other]

  How Can Input Reformulation Improve Tool Usage Accuracy in a Complex Dynamic Environment? A Study on τ-bench 在复杂动态环境中，输入重构如何提高工具使用的准确性？关于 τ -bench 的一项研究Venkatesh Mishra, Amir Saeidi, Satyam Raj, Mutsumi Nakamura, Jayanth Srinivasa, Gaowen Liu, Ali Payani, Chitta Baral Venkatesh Mishra、Amir Saeidi、Satyam Raj、Mutsumi Nakamura、Jayanth Srinivasa、Gaowen Liu、Ali Payani、Chitta BaralComments: Accepted to EMNLP 2025 Findings 评注：已被 EMNLP 2025 Findings 接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [8] arXiv:2508.20916 [pdf, other]

  SageLM: A Multi-aspect and Explainable Large Language Model for Speech Judgement SageLM：用于语音评判的多方面且可解释的大型语言模型Yuan Ge, Junxiang Zhang, Xiaoqian Liu, Bei Li, Xiangnan Ma, Chenglong Wang, Kaiyang Ye, Yangfan Du, Linfeng Zhang, Yuxin Huang, Tong Xiao, Zhengtao Yu, JingBo Zhu 袁戈、张俊翔、刘晓倩、李贝、马向南、王成龙、叶凯阳、杜杨帆、张林峰、黄宇新、肖彤、于正涛、朱靖波Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [9] arXiv:2508.20893 [pdf, html, other]

  The Uneven Impact of Post-Training Quantization in Machine Translation 后训练量化在机器翻译中的不均等影响Benjamin Marie, Atsushi FujitaSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [10] arXiv:2508.20867 [pdf, html, other]

  MSRS: Evaluating Multi-Source Retrieval-Augmented Generation MSRS：评估多源检索增强生成Rohan Phanse, Yijie Zhou, Kejian Shi, Wencai Zhang, Yixin Liu, Yilun Zhao, Arman Cohan Rohan Phanse、周逸杰、施可键、张文才、刘奕欣、赵一伦、Arman CohanComments: COLM 2025; this article supersedes the preprint: arXiv:2309.08960 备注：COLM 2025；本文取代预印本：arXiv:2309.08960Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [11] arXiv:2508.20828 [pdf, html, other]

  GDLLM: A Global Distance-aware Modeling Approach Based on Large Language Models for Event Temporal Relation Extraction GDLLM：一种基于大型语言模型的全局距离感知事件时间关系抽取方法Jie Zhao, Wanting Ning, Yuxiao Fei, Yubo Feng, Lishuang Li 赵杰、宁婉婷、费宇霄、冯雨博、李丽爽Comments: Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP Findings) 备注：2025 年自然语言处理经验方法会议（EMNLP Findings）论文集Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；信息检索 (cs.IR)

• [12] arXiv:2508.20805 [pdf, html, other]

  Exploring Machine Learning and Language Models for Multimodal Depression Detection 探索用于多模态抑郁检测的机器学习与语言模型Javier Si Zhao Hong, Timothy Zoe Delaya, Sherwyn Chan Yin Kit, Pai Chet Ng, Xiaoxiao MiaoComments: This paper has been accepted by APCIPA ASC 2025 备注：本文已被 APCIPA ASC 2025 接收Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Sound (cs.SD) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；声学（cs.SD）

• [13] arXiv:2508.20771 [pdf, html, other]

  Signs of Struggle: Spotting Cognitive Distortions across Language and Register 挣扎的迹象：跨语言与文体识别认知扭曲Abhishek Kuber, Enrico Liscio, Ruixuan Zhang, Caroline Figueroa, Pradeep K. Murukannaiah Abhishek Kuber、Enrico Liscio、Ruixuan Zhang、Caroline Figueroa、Pradeep K. MurukannaiahSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [14] arXiv:2508.20766 [pdf, html, other]

  Turning the Spell Around: Lightweight Alignment Amplification via Rank-One Safety Injection 反转魔咒：通过一阶安全注入实现轻量级对齐放大Harethah Abu Shairah, Hasan Abed Al Kader Hammoud, George Turkiyyah, Bernard Ghanem Harethah Abu Shairah、Hasan Abed Al Kader Hammoud、George Turkiyyah、Bernard GhanemComments: Under Review 评论：审查中Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [15] arXiv:2508.20764 [pdf, html, other]

  Feel the Difference? A Comparative Analysis of Emotional Arcs in Real and LLM-Generated CBT Sessions 感觉不同吗？真实与 LLM 生成 CBT 会话中情感弧线的比较分析Xiaoyi Wang, Jiwei Zhang, Guangtao Zhang, Honglei Guo 王晓毅，张继伟，张光韬，郭红磊Comments: Accepted at EMNLP 2025,14 page,3 figures 评论：已被 EMNLP 2025 接收，14 页，3 张图Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [16] arXiv:2508.20757 [pdf, html, other]

  GUARD: Glocal Uncertainty-Aware Robust Decoding for Effective and Efficient Open-Ended Text Generation GUARD：面向高效且有效的开放式文本生成的全局-局部不确定性感知鲁棒解码Yuanhao Ding, Esteban Garces Arias, Meimingwei Li, Julian Rodemann, Matthias Aßenmacher, Danlu Chen, Gaojuan Fan, Christian Heumann, Chongsheng Zhang 丁元豪，Esteban Garces Arias，李美明伟，Julian Rodemann，Matthias Aßenmacher，陈丹璐，范高娟，Christian Heumann，张崇胜Comments: Accepted at Findings of the Association for Computational Linguistics: EMNLP (Findings) 2025 注释：被接收于计算语言学协会研究成果会议：EMNLP (Findings) 2025Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [17] arXiv:2508.20750 [pdf, html, other]

  Specializing General-purpose LLM Embeddings for Implicit Hate Speech Detection across Datasets 将通用 LLM 嵌入向量专门化以在不同数据集上检测隐式仇恨言论Vassiliy Cheremetiev, Quang Long Ho Ngo, Chau Ying Kot, Alina Elena Baia, Andrea CavallaroComments: Paper accepted at the DHOW Workshop at ACM Multimedia 2025. Code available at this https URL 注释：论文已被 ACM 多媒体 2025 年 DHOW 研讨会接收。代码可在此 https URL 获取Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [18] arXiv:2508.20736 [pdf, html, other]

  Leveraging Semantic Triples for Private Document Generation with Local Differential Privacy Guarantees 利用语义三元组进行私有文档生成并提供本地差分隐私保证Stephen Meisenbacher, Maulik Chevli, Florian MatthesComments: 17 pages, 2 figures, 11 tables. Accepted to EMNLP 2025 (Main) 注释：17 页，2 幅图，11 张表。已被接收至 EMNLP 2025（Main）Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [19] arXiv:2508.20722 [pdf, html, other]

  rStar2-Agent: Agentic Reasoning Technical Report rStar2-Agent：具代理性推理技术报告Ning Shang, Yifei Liu, Yi Zhu, Li Lyna Zhang, Weijiang Xu, Xinyu Guan, Buze Zhang, Bingcheng Dong, Xudong Zhou, Bowen Zhang, Ying Xin, Ziming Miao, Scarlett Li, Fan Yang, Mao YangSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [20] arXiv:2508.20718 [pdf, html, other]

  Addressing Tokenization Inconsistency in Steganography and Watermarking Based on Large Language Models 解决基于大型语言模型的隐写与水印中分词不一致问题Ruiyi Yan, Yugo Murawaki 严瑞益，村脇悠吾Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [21] arXiv:2508.20712 [pdf, html, other]

  Multi-Lingual Implicit Discourse Relation Recognition with Multi-Label Hierarchical Learning 多语种隐式语篇关系识别的多标签层次学习Nelson Filipe Costa, Leila Kosseim Nelson Filipe Costa，Leila KosseimComments: Published at SIGDIAL 2025. Best paper award 评注：发表于 SIGDIAL 2025。最佳论文奖Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [22] arXiv:2508.20700 [pdf, html, other]

  Generative Annotation for ASR Named Entity Correction 用于自动语音识别命名实体纠错的生成式注释Yuanchang Luo, Daimeng Wei, Shaojun Li, Hengchao Shang, Jiaxin Guo, Zongyao Li, Zhanglin Wu, Xiaoyu Chen, Zhiqiang Rao, Jinlong Yang, Hao Yang 罗元昌，魏岱萌，李少军，尚恒超，郭佳欣，李宗尧，吴章林，陈晓宇，饶志强，杨金龙，杨昊Comments: 12 pages, 7 figures, 7 tables, EMNLP 2025 评论：12 页，7 幅图，7 张表，EMNLP 2025Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [23] arXiv:2508.20583 [pdf, html, other]

  A Graph Talks, But Who’s Listening? Rethinking Evaluations for Graph-Language Models 一幅图在讲话，但谁在倾听？重新思考图-语言模型的评估Soham Petkar, Hari Aakash K, Anirudh Vempati, Akshit Sinha, Ponnurangam Kumarauguru, Chirag Agarwal Soham Petkar、Hari Aakash K、Anirudh Vempati、Akshit Sinha、Ponnurangam Kumarauguru、Chirag AgarwalSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [24] arXiv:2508.20567 [pdf, html, other]

  KCS: Diversify Multi-hop Question Generation with Knowledge Composition Sampling KCS：通过知识组合采样多跳问题生成的多样化Yangfan Wang, Jie Liu, Chen Tang, Lian Yan, Jingchi Jiang 王杨帆、刘洁、唐晨、闫璉、蒋景池Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [25] arXiv:2508.20559 [pdf, html, other]

  Leveraging Generative Models for Real-Time Query-Driven Text Summarization in Large-Scale Web Search 在大规模网络搜索中利用生成模型进行实时基于查询的文本摘要Zeyu Xiong, Yixuan Nan, Li Gao, Hengzhu Tang, Shuaiqiang Wang, Junfeng Wang, Dawei Yin 熊泽宇、南奕轩、高丽、唐恒柱、王帅强、王俊峰、尹大为Comments: CIKM'25 备注：CIKM'25Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；信息检索 (cs.IR)

• [26] arXiv:2508.20557 [pdf, html, other]

  Adaptive Federated Distillation for Multi-Domain Non-IID Textual Data 面向多域非独立同分布文本数据的自适应联邦蒸馏Jiahao Xiao, Jiangming Liu 肖嘉豪，刘江明Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [27] arXiv:2508.20554 [pdf, html, other]

  Overview of BioASQ 2025: The Thirteenth BioASQ Challenge on Large-Scale Biomedical Semantic Indexing and Question Answering BioASQ 2025 概述：第十三届大型生物医学语义索引与问答挑战赛Anastasios Nentidis, Georgios Katsimpras, Anastasia Krithara, Martin Krallinger, Miguel Rodríguez-Ortega, Eduard Rodriguez-López, Natalia Loukachevitch, Andrey Sakhovskiy, Elena Tutubalina, Dimitris Dimitriadis, Grigorios Tsoumakas, George Giannakoulas, Alexandra Bekiaridou, Athanasios Samaras, Giorgio Maria Di Nunzio, Nicola Ferro, Stefano Marchesin, Marco Martinelli, Gianmaria Silvello, Georgios Paliouras Anastasios Nentidis、Georgios Katsimpras、Anastasia Krithara、Martin Krallinger、Miguel Rodríguez-Ortega、Eduard Rodriguez-López、Natalia Loukachevitch、Andrey Sakhovskiy、Elena Tutubalina、Dimitris Dimitriadis、Grigorios Tsoumakas、George Giannakoulas、Alexandra Bekiaridou、Athanasios Samaras、Giorgio Maria Di Nunzio、Nicola Ferro、Stefano Marchesin、Marco Martinelli、Gianmaria Silvello、Georgios PaliourasComments: 26 pages, 17 tables, 1 figure 注释：26 页，17 张表，1 张图Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；信息检索 (cs.IR)

• [28] arXiv:2508.20532 [pdf, html, other]

  Overview of BioASQ 2024: The twelfth BioASQ challenge on Large-Scale Biomedical Semantic Indexing and Question Answering BioASQ 2024 概览：第十二届大型生物医学语义索引与问答挑战赛Anastasios Nentidis, Georgios Katsimpras, Anastasia Krithara, Salvador Lima-López, Eulàlia Farré-Maduell, Martin Krallinger, Natalia Loukachevitch, Vera Davydova, Elena Tutubalina, Georgios Paliouras Anastasios Nentidis、Georgios Katsimpras、Anastasia Krithara、Salvador Lima-López、Eulàlia Farré-Maduell、Martin Krallinger、Natalia Loukachevitch、Vera Davydova、Elena Tutubalina、Georgios PaliourasComments: 25 pages, 16 tables, 1 figure 注释：25 页，16 张表，1 张图Journal-ref: Experimental IR Meets Multilinguality, Multimodality, and Interaction. CLEF 2024. Lecture Notes in Computer Science, vol 14959. Springer, Cham 期刊引用：Experimental IR Meets Multilinguality, Multimodality, and Interaction。CLEF 2024。Lecture Notes in Computer Science，卷 14959。Springer，ChamSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；信息检索 (cs.IR)

• [29] arXiv:2508.20514 [pdf, html, other]

  SciTopic: Enhancing Topic Discovery in Scientific Literature through Advanced LLM SciTopic：通过先进的 LLM 增强学术文献主题发现Pengjiang Li, Zaitian Wang, Xinhao Zhang, Ran Zhang, Lu Jiang, Pengfei Wang, Yuanchun Zhou 李鹏江, 王在天, 张欣豪, 张然, 江璐, 王鹏飞, 周元春Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [30] arXiv:2508.20511 [pdf, html, other]

  Languages Still Left Behind: Toward a Better Multilingual Machine Translation Benchmark 仍被忽视的语言：迈向更好的多语言机器翻译基准Chihiro Taguchi, Seng Mai, Keita Kurabe, Yusuke Sakai, Georgina Agyei, Soudabeh Eslami, David Chiang 田口千尋, Seng Mai, 倉部恵太, 酒井祐介, Georgina Agyei, Soudabeh Eslami, David ChiangComments: 13 pages, 7 tables, 2 figures. Accepted at EMNLP Main 2025. Code and data released at this https URL 备注：13 页，7 表，2 图。被接受于 EMNLP Main 2025。代码和数据已在此 https URL 发布Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [31] arXiv:2508.20468 [pdf, other]

  ConspirED: A Dataset for Cognitive Traits of Conspiracy Theories and Large Language Model Safety ConspirED：关于阴谋论的认知特征与大语言模型安全的数据集Luke Bates, Max Glockner, Preslav Nakov, Iryna GurevychSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [32] arXiv:2508.20460 [pdf, html, other]

  Prediction of mortality and resource utilization in critical care: a deep learning approach using multimodal electronic health records with natural language processing techniques 使用自然语言处理技术对多模态电子病历进行深度学习以预测重症监护中的死亡率和资源利用Yucheng Ruan, Xiang Lan, Daniel J. Tan, Hairil Rizal Abdullah, Mengling Feng 阮羽成，兰翔，丹尼尔·J·谭，海里尔·里扎尔·阿卜杜拉，冯孟玲Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [33] arXiv:2508.20453 [pdf, other]

  MCP-Bench: Benchmarking Tool-Using LLM Agents with Complex Real-World Tasks via MCP Servers MCP-Bench：通过 MCP 服务器使用 LLM 代理对复杂现实任务进行基准测试的工具Zhenting Wang, Qi Chang, Hemani Patel, Shashank Biju, Cheng-En Wu, Quan Liu, Aolin Ding, Alireza Rezazadeh, Ankit Shah, Yujia Bao, Eugene Siow 王振庭，常琦，赫玛尼·帕特尔，沙尚克·比珠，陈恩武，刘泉，丁奥林，阿里雷扎·雷扎扎德，安基特·沙，鲍钰佳，尤金·肖Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [34] arXiv:2508.20442 [pdf, other]

  Searching the Title of Practical Work of the Informatics Engineering Bachelor Program with the Case Base Reasoning Method 使用个案推理方法检索信息工程学士课程实践工作题目Agung Sukrisna Jaya, Osvari Arsalan, Danny Matthew Saputra 阿贡·苏克里斯纳·贾亚，奥斯瓦里·阿尔萨兰，丹尼·马修·萨普特拉Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [35] arXiv:2508.20420 [pdf, html, other]

  CAMB: A comprehensive industrial LLM benchmark on civil aviation maintenance CAMB：一个关于民航维修的工业级 LLM 全面基准测试Feng Zhang, Chengjie Pang, Yuehan Zhang, Chenyu Luo 张风，庞成杰，张悦涵，罗辰宇Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [36] arXiv:2508.20417 [pdf, html, other]

  KG-CQR: Leveraging Structured Relation Representations in Knowledge Graphs for Contextual Query Retrieval KG-CQR：在知识图谱中利用结构化关系表示进行上下文查询检索Chi Minh Bui, Ngoc Mai Thieu, Van Vinh Nguyen, Json J.Jung, Khac-Hoai Nam Bui Chi Minh Bui、Ngoc Mai Thieu、Van Vinh Nguyen、Json J.Jung、Khac-Hoai Nam BuiComments: Accepted at Main EMNLP 2025 评论：已被接受于 EMNLP 2025 主会Subjects: Computation and Language (cs.CL); Databases (cs.DB) 学科：计算与语言（cs.CL）；数据库（cs.DB）

• [37] arXiv:2508.20416 [pdf, html, other]

  DentalBench: Benchmarking and Advancing LLMs Capability for Bilingual Dentistry Understanding DentalBench：评估并推进 LLMs 在双语牙科理解能力方面的基准测试Hengchuan Zhu, Yihuan Xu, Yichen Li, Zijie Meng, Zuozhu Liu 朱恒川，徐艺桓，李意辰，孟子杰，刘作铸Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [38] arXiv:2508.20410 [pdf, other]

  UI-Bench: A Benchmark for Evaluating Design Capabilities of AI Text-to-App Tools UI-Bench：用于评估 AI 文本到应用工具设计能力的基准Sam Jung, Agustin Garcinuno, Spencer Mateega Sam Jung、Agustin Garcinuno、Spencer MateegaSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [39] arXiv:2508.20395 [pdf, html, other]

  Measuring Reasoning Utility in LLMs via Conditional Entropy Reduction 通过条件熵减少衡量 LLMs 的推理效用Xu Guo 郭旭Comments: 11 pages, 4 figures 注：11 页，4 幅图Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [40] arXiv:2508.20385 [pdf, html, other]

  CAPE: Context-Aware Personality Evaluation Framework for Large Language Models CAPE：面向上下文的语言模型人格评估框架Jivnesh Sandhan, Fei Cheng, Tushar Sandhan, Yugo Murawaki Jivnesh Sandhan、Fei Cheng、Tushar Sandhan、Yugo MurawakiComments: Accepted at EMNLP25 (Findings) 备注：已被 EMNLP25（Findings）接受Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [41] arXiv:2508.20373 [pdf, other] [41] arXiv:2508.20373 [ pdf，其他]

  Graph-R1: Unleashing LLM Reasoning with NP-Hard Graph Problems Graph-R1：通过 NP-困难图问题释放 LLM 推理能力Yuyao Wang, Bowen Liu, Jianheng Tang, Nuo Chen, Yuhan Li, Qifan Zhang, Jia Li 王禹尧、刘博文、唐建恒、陈诺、李雨涵、张启凡、李佳Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [42] arXiv:2508.20351 [pdf, html, other]

  Joint Enhancement of Relational Reasoning for Long-Context LLMs 面向长上下文 LLMs 的关系推理联合增强Zhirui Chen, Wei Shen, Jiashui Huang, Ling Shao 陈志锐、沈蔚、黄家随、沙凌Comments: 9 pages, 5 pages Accepted by EMNLP 2025 Findings 注释：9 页，5 页 被 EMNLP 2025 Findings 接受Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [43] arXiv:2508.20325 [pdf, other]

  GUARD: Guideline Upholding Test through Adaptive Role-play and Jailbreak Diagnostics for LLMs GUARD：通过自适应角色扮演与越狱诊断验证对 LLMs 的准则遵守Haibo Jin, Ruoxi Chen, Peiyan Zhang, Andy Zhou, Yang Zhang, Haohan Wang 金海博、陈若曦、张佩彦、周安迪、张扬、王昊涵Comments: 54 pages 注释：54 页Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；计算机视觉与模式识别 (cs.CV)

• [44] arXiv:2508.20324 [pdf, html, other]

  Can Compact Language Models Search Like Agents? Distillation-Guided Policy Optimization for Preserving Agentic RAG Capabilities 紧凑型语言模型能像代理一样进行搜索吗？蒸馏引导的策略优化以保留具代理性的 RAG 能力Rikuto Kotoge, Mai Nishimura, Jiaxin Ma Kotoge Rikuto、Nishimura Mai、Ma JiaxinSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [45] arXiv:2508.20223 [pdf, html, other] [45] arXiv:2508.20223 [ pdf，html，other]

  Integrating SystemC TLM into FMI 3.0 Co-Simulations with an Open-Source Approach 将 SystemC TLM 集成到 FMI 3.0 协同仿真中的开源方法Andrei Mihai Albu, Giovanni Pollo, Alessio Burrello, Daniele Jahier Pagliari, Cristian Tesconi, Alessandra Neri, Dario Soldi, Fabio Autieri, Sara Vinco Andrei Mihai Albu、Giovanni Pollo、Alessio Burrello、Daniele Jahier Pagliari、Cristian Tesconi、Alessandra Neri、Dario Soldi、Fabio Autieri、Sara VincoSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [46] arXiv:2508.20217 [pdf, other] [46] arXiv:2508.20217 [ pdf，other]

  Prompting Strategies for Language Model-Based Item Generation in K-12 Education: Bridging the Gap Between Small and Large Language Models 面向 K-12 教育的基于语言模型的试题生成提示策略：弥合小型与大型语言模型之间的差距Mohammad Amini, Babak Ahmadi, Xiaomeng Xiong, Yilin Zhang, Christopher Qiao Mohammad Amini、Babak Ahmadi、Xiaomeng Xiong、Yilin Zhang、Christopher QiaoSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [47] arXiv:2508.20201 [pdf, html, other]

  Social Bias in Multilingual Language Models: A Survey 多语言语言模型中的社会偏见：综述Lance Calvin Lim Gamboa, Yue Feng, Mark Lee 兰斯·卡尔文·林·甘博亚、岳锋、马克·李Comments: Accepted into EMNLP 2025 Main Conference 评审意见：已被接收至 EMNLP 2025 主会场Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [48] arXiv:2508.21038 (cross-list from cs.IR) [pdf, html, other] [48] arXiv:2508.21038（来自 cs.IR 的跨列表）[ pdf, html, other]

  On the Theoretical Limitations of Embedding-Based Retrieval 关于基于嵌入的检索的理论局限性Orion Weller, Michael Boratko, Iftekhar Naim, Jinhyuk LeeSubjects: Information Retrieval (cs.IR); Computation and Language (cs.CL); Machine Learning (cs.LG) 学科：信息检索（cs.IR）；计算与语言（cs.CL）；机器学习（cs.LG）

• [49] arXiv:2508.21010 (cross-list from cs.CV) [pdf, html, other] [49] arXiv:2508.21010（交叉分类自 cs.CV）[ pdf, html, other]

  ChainReaction! Structured Approach with Causal Chains as Intermediate Representations for Improved and Explainable Causal Video Question Answering ChainReaction！以因果链作为中间表示的结构化方法，用于改进且可解释的因果视频问答Paritosh Parmar, Eric Peh, Basura Fernando Paritosh Parmar、Eric Peh、Basura FernandoComments: Project page: this https URL 备注：项目页面：this https URLSubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG) 学科：计算机视觉与模式识别（cs.CV）；人工智能（cs.AI）；计算与语言（cs.CL）；人机交互（cs.HC）；机器学习（cs.LG）

• [50] arXiv:2508.20869 (cross-list from cs.SD) [pdf, html, other] [50] arXiv:2508.20869（从 cs.SD 交叉列出）[ pdf, html, other]

  OLMoASR: Open Models and Data for Training Robust Speech Recognition Models OLMoASR：用于训练鲁棒语音识别模型的开放模型和数据Huong Ngo, Matt Deitke, Martijn Bartelds, Sarah Pratt, Josh Gardner, Matt Jordan, Ludwig SchmidtComments: 17 pages, 7 figures 备注：17 页，7 张图Subjects: Sound (cs.SD); Computation and Language (cs.CL); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS) 主题：声音 (cs.SD)；计算与语言 (cs.CL)；机器学习 (cs.LG)；音频与语音处理 (eess.AS)

• [51] arXiv:2508.20810 (cross-list from cs.AI) [pdf, html, other] [51] arXiv:2508.20810（跨列自 cs.AI）[ pdf, html, other]

  A Graph-Based Test-Harness for LLM Evaluation 基于图的 LLM 评估测试平台Jessica Lundin, Guillaume Chabot-CoutureComments: 4 pages, 2 figures, dataset 备注：4 页，2 幅图，数据集Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [52] arXiv:2508.20701 (cross-list from cs.AI) [pdf, html, other] [52] arXiv:2508.20701（来自 cs.AI 的交叉列表）[ pdf, html, other]

  Transparent Semantic Spaces: A Categorical Approach to Explainable Word Embeddings 透明语义空间：一种基于范畴论的可解释词嵌入方法Ares Fabregat-Hernández (1 and 2), Javier Palanca (1), Vicent Botti (1 and 3) ((1) Valencian Research Institute for Artificial Intelligence (VRAIN) Universitat Politècnica de València (2) Universidad Internacional de Valencia (VIU) (3) valgrAI (Valencian Graduate School and Research Network of Artificial Intelligence)) Ares Fabregat-Hernández (1 和 2)、Javier Palanca (1)、Vicent Botti (1 和 3)（(1) 瓦伦西亚人工智能研究所（VRAIN） 瓦伦西亚理工大学 (2) 瓦伦西亚国际大学（VIU） (3) valgrAI（瓦伦西亚人工智能研究生院与研究网络））Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Category Theory (math.CT) 主题：人工智能 (cs.AI); 计算与语言 (cs.CL); 范畴论 (math.CT)

• [53] arXiv:2508.20697 (cross-list from cs.LG) [pdf, html, other] [53] arXiv:2508.20697（从 cs.LG 交叉列出）[ pdf, html, other]

  Token Buncher: Shielding LLMs from Harmful Reinforcement Learning Fine-Tuning Token Buncher：保护 LLMs 免受有害的强化学习微调影响Weitao Feng, Lixu Wang, Tianyi Wei, Jie Zhang, Chongyang Gao, Sinong Zhan, Peizhuo Lv, Wei DongComments: Project Hompage: this https URL 评注：项目主页：这个 https URLSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL) 学科：机器学习（cs.LG）；计算与语言（cs.CL）

• [54] arXiv:2508.20693 (cross-list from cs.DL) [pdf, html, other] [54] arXiv:2508.20693（从 cs.DL 交叉列出）[ pdf, html, other]

  Leveraging Large Language Models for Generating Research Topic Ontologies: A Multi-Disciplinary Study 利用大型语言模型生成研究主题本体：一项多学科研究Tanay Aggarwal, Angelo Salatino, Francesco Osborne, Enrico Motta Tanay Aggarwal、Angelo Salatino、Francesco Osborne、Enrico MottaSubjects: Digital Libraries (cs.DL); Computation and Language (cs.CL) 主题：数字图书馆（cs.DL）；计算与语言（cs.CL）

• [55] arXiv:2508.20691 (cross-list from cs.CV) [pdf, html, other] [55] arXiv:2508.20691（从 cs.CV 交叉列出）[ pdf，html，other]

  MobileCLIP2: Improving Multi-Modal Reinforced Training MobileCLIP2：改进多模态强化训练Fartash Faghri, Pavan Kumar Anasosalu Vasu, Cem Koc, Vaishaal Shankar, Alexander Toshev, Oncel Tuzel, Hadi Pouransari Fartash Faghri、Pavan Kumar Anasosalu Vasu、Cem Koc、Vaishaal Shankar、Alexander Toshev、Oncel Tuzel、Hadi PouransariComments: TMLR August 2025 评论：TMLR 2025 年 8 月Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算机视觉与模式识别（cs.CV）；人工智能（cs.AI）；计算与语言（cs.CL）；机器学习（cs.LG）

• [56] arXiv:2508.20655 (cross-list from cs.CV) [pdf, html, other] [56] arXiv:2508.20655（从 cs.CV 交叉列出）[ pdf, html, other]

  Improving Alignment in LVLMs with Debiased Self-Judgment 在大规模视觉语言模型中通过去偏自我评估改进对齐性Sihan Yang, Chenhang Cui, Zihao Zhao, Yiyang Zhou, Weilong Yan, Ying Wei, Huaxiu YaoComments: EMNLP 2025 Findings 备注：EMNLP 2025 FindingsSubjects: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL) 主题：计算机视觉与模式识别 (cs.CV)；计算与语言 (cs.CL)

• [57] arXiv:2508.20637 (cross-list from cs.LG) [pdf, html, other] [57] arXiv:2508.20637（跨列表自 cs.LG）[ pdf，html，其他]

  GDS Agent: A Graph Algorithmic Reasoning Agent GDS Agent：一种图算法推理代理Borun Shi, Ioannis Panagiotas Borun Shi，Ioannis PanagiotasComments: Technical report 注释：技术报告Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：机器学习 (cs.LG); 人工智能 (cs.AI); 计算与语言 (cs.CL)

• [58] arXiv:2508.20577 (cross-list from cs.LG) [pdf, html, other] [58] arXiv:2508.20577（来自 cs.LG 的交叉列表）[ pdf, html, other]

  MERIT: Maximum-normalized Element-wise Ratio for Language Model Large-batch Training MERIT：用于语言模型大批量训练的最大归一化逐元素比率Yang Luo, Zangwei Zheng, Ziheng Qin, Zirui Zhu, Yong Liu, Yang You 杨洛，郑藏威，秦子恒，朱子睿，刘勇，游扬Comments: ICML 2025 备注：ICML 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：机器学习 (cs.LG); 人工智能 (cs.AI); 计算与语言 (cs.CL)

• [59] arXiv:2508.20474 (cross-list from eess.AS) [pdf, html, other] [59] arXiv:2508.20474（从 eess.AS 跨列表）[ pdf, html, other]

  Unifying Diarization, Separation, and ASR with Multi-Speaker Encoder 统一说话人分离、分离与语音识别的多说话人编码器Muhammad Shakeel, Yui Sudo, Yifan Peng, Chyi-Jiunn Lin, Shinji WatanabeComments: Accepted to IEEE ASRU 2025 备注：已被 IEEE ASRU 2025 接收Subjects: Audio and Speech Processing (eess.AS); Computation and Language (cs.CL); Sound (cs.SD) 主题：音频与语音处理 (eess.AS); 计算与语言 (cs.CL); 声音 (cs.SD)

• [60] arXiv:2508.20353 (cross-list from cs.LG) [pdf, html, other] [60] arXiv:2508.20353（从 cs.LG 交叉列出）[ pdf, html, other]

  DFAMS: Dynamic-flow guided Federated Alignment based Multi-prototype Search DFAMS：基于动态流引导联邦对齐的多原型搜索Zhibang Yang, Xinke Jiang, Rihong Qiu, Ruiqing Li, Yihang Zhang, Yue Fang, Yongxin Xu, Hongxin Ding, Xu Chu, Junfeng Zhao, Yasha Wang 杨志邦、蒋新科、邱日鸿、李锐青、张逸航、方悦、许永鑫、丁洪欣、褚旭、赵俊峰、王雅莎Comments: 7 pages, 3 figures 注释：7 页，3 幅图Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL) 学科：机器学习（cs.LG）；计算与语言（cs.CL）

• [61] arXiv:2508.20333 (cross-list from cs.LG) [pdf, other] [61] arXiv:2508.20333（从 cs.LG 交叉列出）[ pdf，其他]

  Poison Once, Refuse Forever: Weaponizing Alignment for Injecting Bias in LLMs Poison Once, Refuse Forever：将对齐武器化以向 LLMs 注入偏见Md Abdullah Al Mamun, Ihsen Alouani, Nael Abu-Ghazaleh Md Abdullah Al Mamun、Ihsen Alouani、Nael Abu-GhazalehSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Distributed, Parallel, and Cluster Computing (cs.DC) 学科：机器学习 (cs.LG)；人工智能 (cs.AI)；计算与语言 (cs.CL)；分布式、并行与集群计算 (cs.DC)

• [62] arXiv:2508.20312 (cross-list from cs.IR) [pdf, html, other] [62] arXiv:2508.20312（从 cs.IR 交叉列出）[ pdf，html，other]

  ELIXIR: Efficient and LIghtweight model for eXplaIning Recommendations ELIXIR：用于解释推荐的高效轻量模型（Efficient and LIghtweight model for eXplaIning Recommendations）Ben Kabongo, Vincent Guigue, Pirmin Lemberger 本·卡邦戈、文森特·吉格、皮尔明·莱姆伯格Comments: 10 pages, 3 figures, 6 Tables 备注：10 页，3 幅图，6 个表Subjects: Information Retrieval (cs.IR); Computation and Language (cs.CL); Machine Learning (cs.LG) 学科：信息检索（cs.IR）；计算与语言（cs.CL）；机器学习（cs.LG）

• [63] arXiv:2508.20279 (cross-list from cs.CV) [pdf, html, other] [63] arXiv:2508.20279（从 cs.CV 交叉列出）[ pdf, html, other]

  How Multimodal LLMs Solve Image Tasks: A Lens on Visual Grounding, Task Reasoning, and Answer Decoding 多模态 LLMs 如何解决图像任务：关于视觉定位、任务推理与答案解码的透视Zhuoran Yu, Yong Jae Lee 余卓然，李永在Comments: Accepted by COLM 2025 备注：被 COLM 2025 接收Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：计算机视觉与模式识别（cs.CV）；人工智能（cs.AI）；计算与语言（cs.CL）

• [64] arXiv:2508.20275 (cross-list from cs.LG) [pdf, html, other] [64] arXiv:2508.20275（从 cs.LG 交叉列出）[ pdf, html, other]

  A Systematic Review on the Generative AI Applications in Human Medical Genomics 关于生成式人工智能在人类医学基因组学应用的系统综述Anton Changalidis, Yury Barbitoff, Yulia Nasykhova, Andrey Glotov Anton Changalidis、Yury Barbitoff、Yulia Nasykhova、Andrey GlotovComments: 31 pages, 5 figures 评注：31 页，5 幅图Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL); Quantitative Methods (q-bio.QM) 主题：机器学习 (cs.LG)；计算与语言 (cs.CL)；定量方法 (q-bio.QM)

• [65] arXiv:2508.20228 (cross-list from cs.CR) [pdf, html, other] [65] arXiv:2508.20228（从 cs.CR 交叉列出）[ pdf, html, other]

  Robustness Assessment and Enhancement of Text Watermarking for Google’s SynthID 对谷歌 SynthID 文本水印的鲁棒性评估与增强Xia Han, Qi Li, Jianbing Ni, Mohammad Zulkernine 韩霞，李琦，倪健兵，Mohammad ZulkernineComments: submitted to TrustCom2025 备注：已提交至 TrustCom2025Subjects: Cryptography and Security (cs.CR); Computation and Language (cs.CL) 主题：密码学与安全（cs.CR）；计算与语言（cs.CL）

• [66] arXiv:2508.20227 (cross-list from cs.CV) [pdf, other] [66] arXiv:2508.20227（来自 cs.CV 的交叉列表）[ pdf，其他]

  A Novel Framework for Automated Explain Vision Model Using Vision-Language Models 用于使用视觉-语言模型自动解释视觉模型的新框架Phu-Vinh Nguyen, Tan-Hanh Pham, Chris Ngo, Truong Son Hy Phu-Vinh Nguyen，Tan-Hanh Pham，Chris Ngo，Truong Son HySubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算机视觉与模式识别（cs.CV）；人工智能（cs.AI）；计算与语言（cs.CL）；机器学习（cs.LG）

• [67] arXiv:2508.20195 (cross-list from cs.AI) [pdf, other] [67] arXiv:2508.20195（从 cs.AI 交叉列出）[ pdf，其他]

  AI-AI Esthetic Collaboration with Explicit Semiotic Awareness and Emergent Grammar Development 具有显式符号意识和新生语法发展之人工智能—人工智能美学协作Nicanor I. MoldovanComments: 13 pages 备注：13 页Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multiagent Systems (cs.MA) 主题：人工智能 (cs.AI); 计算与语言 (cs.CL); 多智能体系统 (cs.MA)

• [68] arXiv:2508.20181 (cross-list from cs.CV) [pdf, html, other] [68] arXiv:2508.20181（从 cs.CV 交叉归类）[ pdf, html, other]

  Mitigating Hallucinations in Multimodal LLMs via Object-aware Preference Optimization 通过面向对象的偏好优化缓解多模态 LLMs 的幻觉问题Alberto Compagnoni, Davide Caffagni, Nicholas Moratelli, Lorenzo Baraldi, Marcella Cornia, Rita CucchiaraComments: BMVC 2025 评注：BMVC 2025Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multimedia (cs.MM) 主题：计算机视觉与模式识别 (cs.CV)；人工智能 (cs.AI)；计算与语言 (cs.CL)；多媒体 (cs.MM)

• [69] arXiv:2508.20109 (cross-list from q-bio.NC) [pdf, other] [69] arXiv:2508.20109（从 q-bio.NC 交叉列出）[ pdf，其他]

  A Unified Theory of Language 语言的统一理论Robert Worden 罗伯特·沃登Comments: 54 pages 注释：54 页Subjects: Neurons and Cognition (q-bio.NC); Computation and Language (cs.CL) 主题：神经元与认知（q-bio.NC）；计算与语言（cs.CL）

• [70] arXiv:2508.20068 [pdf, html, other]

  11Plus-Bench: Demystifying Multimodal LLM Spatial Reasoning with Cognitive-Inspired Analysis 11Plus-Bench：以认知启发分析揭开多模态 LLM 空间推理的面纱Chengzu Li, Wenshan Wu, Huanyu Zhang, Qingtao Li, Zeyu Gao, Yan Xia, José Hernández-Orallo, Ivan Vulić, Furu WeiComments: 9 pages, 4 figures (22 pages, 7 figures, 7 tables including references and appendices) 注释：9 页，4 幅图（22 页，7 幅图，7 张表，含参考文献与附录）Subjects: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG) 学科：计算与语言（cs.CL）；计算机视觉与模式识别（cs.CV）；机器学习（cs.LG）

• [71] arXiv:2508.20047 [pdf, html, other] [71] arXiv:2508.20047 [ pdf、html、其他]

  AraHealthQA 2025: The First Shared Task on Arabic Health Question Answering AraHealthQA 2025：首届阿拉伯语健康问答共享任务Hassan Alhuzali, Farah Shamout, Muhammad Abdul-Mageed, Chaimae Abouzahir, Mouath Abu-Daoud, Ashwag Alasmari, Walid Al-Eisawi, Renad Al-Monef, Ali Alqahtani, Lama Ayash, Nizar Habash, Leen Kharouf 哈桑·阿尔胡扎利、法拉·沙穆特、穆罕默德·阿卜杜勒-马吉德、查伊玛·阿布扎希尔、穆阿兹·阿布-达乌德、阿什瓦格·阿拉斯马里、瓦利德·阿尔-艾萨维、雷纳德·阿尔-莫内夫、阿里·阿尔卡赫塔尼、拉玛·阿亚什、尼扎尔·哈巴什、琳·哈鲁夫Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [72] arXiv:2508.20038 [pdf, html, other]

  Forewarned is Forearmed: Pre-Synthesizing Jailbreak-like Instructions to Enhance LLM Safety Guardrail to Potential Attacks 先发制人：预先合成类越狱指令以增强 LLM 对潜在攻击的安全防护栏Sheng Liu, Qiang Sheng, Danding Wang, Yang Li, Guang Yang, Juan Cao 盛柳、盛强、王丹丁、李洋、杨广、曹娟Comments: EMNLP 2025 findings 评论：EMNLP 2025 发现Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [73] arXiv:2508.20033 [pdf, html, other]

  DeepScholar-Bench: A Live Benchmark and Automated Evaluation for Generative Research Synthesis DeepScholar-Bench：用于生成式研究综述的实时基准与自动评估Liana Patel, Negar Arabzadeh, Harshit Gupta, Ankita Sundar, Ion Stoica, Matei Zaharia, Carlos GuestrinSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [74] arXiv:2508.19997 [pdf, html, other]

  Selective Retrieval-Augmentation for Long-Tail Legal Text Classification 面向长尾法律文本分类的选择性检索增强Boheng Mao 茅博恒Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；信息检索 (cs.IR)

• [75] arXiv:2508.19996 [pdf, html, other]

  ReSURE: Regularizing Supervision Unreliability for Multi-turn Dialogue Fine-tuning ReSURE：针对多轮对话微调的监督不可靠性正则化Yiming Du, Yifan Xiang, Bin Liang, Dahua Lin, Kam-Fai Wong, Fei Tan 杜奕铭，向一凡，梁斌，林达华，黄锦辉，谭飞Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [76] arXiv:2508.19993 [pdf, html, other]

  MathBuddy: A Multimodal System for Affective Math Tutoring MathBuddy：一个用于情感化数学辅导的多模态系统Debanjana Kar, Leopold Böss, Dacia Braca, Sebastian Maximilian Dennerlein, Nina Christine Hubig, Philipp Wintersberger, Yufang Hou Debanjana Kar、Leopold Böss、Dacia Braca、Sebastian Maximilian Dennerlein、Nina Christine Hubig、Philipp Wintersberger、Yufang HouSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；人机交互 (cs.HC)

• [77] arXiv:2508.19988 [pdf, other]

  AgentCoMa: A Compositional Benchmark Mixing Commonsense and Mathematical Reasoning in Real-World Scenarios AgentCoMa：一个将常识与数学推理混合于现实场景的组合性基准Lisa Alazraki, Lihu Chen, Ana Brassard, Joe Stacey, Hossein A. Rahmani, Marek Rei Lisa Alazraki、Lihu Chen、Ana Brassard、Joe Stacey、Hossein A. Rahmani、Marek ReiSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [78] arXiv:2508.19982 [pdf, html, other]

  Diffusion Language Models Know the Answer Before Decoding 扩散语言模型在解码前就已知答案Pengxiang Li, Yefan Zhou, Dilxat Muhtar, Lu Yin, Shilin Yan, Li Shen, Yi Liang, Soroush Vosoughi, Shiwei Liu 李鹏翔，周叶凡，Dilxat Muhtar，尹璐，颜世琳，沈力，梁毅，Soroush Vosoughi，刘士炜Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [79] arXiv:2508.19966 [pdf, html, other]

  Dhati+: Fine-tuned Large Language Models for Arabic Subjectivity Evaluation Dhati+：为阿拉伯语主观性评估微调的大型语言模型Slimane Bellaouar, Attia Nehar, Soumia Souffi, Mounia Bouameur Slimane Bellaouar、Attia Nehar、Soumia Souffi、Mounia BouameurComments: 25 pages, 7 figures 备注：25 页，7 幅图Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [80] arXiv:2508.19922 [pdf, html, other]

  HEAL: A Hypothesis-Based Preference-Aware Analysis Framework HEAL：一个基于假设的偏好感知分析框架Yifu Huo, Chenglong Wang, Qiren Zhu, Shunjie Xing, Tong Xiao, Chunliang Zhang, Tongran Liu, Jinbo Zhu 霍艺甫，王成龙，朱启人，邢舜捷，肖彤，张春良，刘通然，朱晋博Comments: Accepted by EMNLP 2025 Findings 备注：被 EMNLP 2025 Findings 接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [81] arXiv:2508.19919 [pdf, html, other]

  Your AI Bosses Are Still Prejudiced: The Emergence of Stereotypes in LLM-Based Multi-Agent Systems 你的 AI 上司仍然存在偏见：基于 LLM 的多智能体系统中刻板印象的出现Jingyu Guo, Yingying Xu 郭靖宇，徐莹莹Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [82] arXiv:2508.19903 [pdf, html, other]

  Logical Reasoning with Outcome Reward Models for Test-Time Scaling 用于测试时扩展的带有结果奖励模型的逻辑推理Ramya Keerthy Thatikonda, Wray Buntine, Ehsan ShareghiComments: EMNLP 2025 评论：EMNLP 2025Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [83] arXiv:2508.19887 [pdf, other]

  Bangla-Bayanno: A 52K-Pair Bengali Visual Question Answering Dataset with LLM-Assisted Translation Refinement Bangla-Bayanno：一个包含 52K 对孟加拉语视觉问答的数据集，使用 LLM 辅助的翻译精炼Mohammed Rakibul Hasan, Rafi Majid, Ahanaf Tahmid 穆罕默德·拉基布尔·哈桑，拉菲·马吉德，阿哈纳夫·塔赫米德Subjects: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV) 学科：计算与语言 (cs.CL)；计算机视觉与模式识别 (cs.CV)

• [84] arXiv:2508.19883 [pdf, other]

  AI-Powered Detection of Inappropriate Language in Medical School Curricula 利用人工智能检测医学院课程中不当语言Chiman Salavati, Shannon Song, Scott A. Hale, Roberto E. Montenegro, Shiri Dori-Hacohen, Fabricio Murai Chiman Salavati、Shannon Song、Scott A. Hale、Roberto E. Montenegro、Shiri Dori-Hacohen、Fabricio MuraiComments: Accepted at 2025 AAAI/ACM AI, Ethics and Society Conference (AIES'25) 评审意见：被 2025 年 AAAI/ACM 人工智能、伦理与社会会议（AIES'25）接受Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；计算机与社会 (cs.CY)

• [85] arXiv:2508.19873 [pdf, html, other]

  Beyond Shallow Heuristics: Leveraging Human Intuition for Curriculum Learning 超越浅层启发式：利用人类直觉进行课程学习Vanessa Toborek, Sebastian Müller, Tim Selbach, Tamás Horváth, Christian Bauckhage Vanessa Toborek、Sebastian Müller、Tim Selbach、Tamás Horváth、Christian BauckhageComments: Presented at ICNLSP 2025; to appear in the ACL Anthology; received the Best Short Paper Award 备注：在 ICNLSP 2025 上报告；将收录于 ACL Anthology；获得最佳短文奖Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [86] arXiv:2508.19856 [pdf, html, other]

  TokenVerse++: Towards Flexible Multitask Learning with Dynamic Task Activation TokenVerse++：走向具有动态任务激活的灵活多任务学习Shashi Kumar, Srikanth Madikeri, Esaú Villatoro-Tello, Sergio Burdisso, Pradeep Rangappa, Andrés Carofilis, Petr Motlicek, Karthik Pandia, Shankar Venkatesan, Kadri Hacioğlu, Andreas Stolcke Shashi Kumar、Srikanth Madikeri、Esaú Villatoro-Tello、Sergio Burdisso、Pradeep Rangappa、Andrés Carofilis、Petr Motlicek、Karthik Pandia、Shankar Venkatesan、Kadri Hacioğlu、Andreas StolckeComments: Accepted to IEEE ASRU 2025. Copyright©2025 IEEE 备注：已被 IEEE ASRU 2025 接收。版权所有©2025 IEEESubjects: Computation and Language (cs.CL); Audio and Speech Processing (eess.AS) 主题：计算与语言（cs.CL）；音频与语音处理（eess.AS）

• [87] arXiv:2508.19836 [pdf, html, other]

  Scalable and consistent few-shot classification of survey responses using text embeddings 使用文本嵌入进行可扩展且一致的少样本调查回复分类Jonas Timmann Mjaaland, Markus Fleten Kreutzer, Halvor Tyseng, Rebeckah K. Fussell, Gina Passante, N.G. Holmes, Anders Malthe-Sørenssen, Tor Ole B. OddenSubjects: Computation and Language (cs.CL); Physics Education (physics.ed-ph) 学科：计算与语言 (cs.CL)；物理教育 (physics.ed-ph)

• [88] arXiv:2508.19831 [pdf, html, other]

  Benchmarking Hindi LLMs: A New Suite of Datasets and a Comparative Analysis 基准测试印地语 LLMs：一套新的数据集与比较分析Anusha Kamath, Kanishk Singla, Rakesh Paul, Raviraj Joshi, Utkarsh Vaidya, Sanjay Singh Chauhan, Niranjan WartikarSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [89] arXiv:2508.19828 [pdf, html, other]

  Memory-R1: Enhancing Large Language Model Agents to Manage and Utilize Memories via Reinforcement Learning Memory-R1：通过强化学习增强大型语言模型代理以管理和利用记忆Sikuan Yan, Xiufeng Yang, Zuchao Huang, Ercong Nie, Zifeng Ding, Zonggen Li, Xiaowen Ma, Hinrich Schütze, Volker Tresp, Yunpu Ma 闫思宽，杨秀峰，黄祖超，聂二聪，丁子峰，李纵根，马晓文，希特策（Hinrich Schütze），沃尔克·特雷斯普（Volker Tresp），马云谱Subjects: Computation and Language (cs.CL); Multiagent Systems (cs.MA) 主题：计算与语言（cs.CL）；多智能体系统（cs.MA）

• [90] arXiv:2508.19813 [pdf, other]

  T2R-bench: A Benchmark for Generating Article-Level Reports from Real World Industrial Tables T2R-bench：一个用于从真实工业表格生成整篇文章级报告的基准测试Jie Zhang, Changzai Pan, Kaiwen Wei, Sishi Xiong, Yu Zhao, Xiangyu Li, Jiaxin Peng, Xiaoyan Gu, Jian Yang, Wenhan Chang, Zhenhe Wu, Jiang Zhong, Shuangyong Song, Yongxiang Li, Xuelong LiSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [91] arXiv:2508.19764 [pdf, other]

  Principled Personas: Defining and Measuring the Intended Effects of Persona Prompting on Task Performance 有原则的人设：定义并衡量人设提示对任务表现的预期效果Pedro Henrique Luz de Araujo, Paul Röttger, Dirk Hovy, Benjamin RothComments: 30 pages, 29 figures, accepted to EMNLP 2025 评论：30 页，29 幅图，已被 EMNLP 2025 接受Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [92] arXiv:2508.19758 [pdf, html, other]

  Uncovering the Bigger Picture: Comprehensive Event Understanding Via Diverse News Retrieval 揭示更大的图景：通过多样化新闻检索实现全面的事件理解Yixuan Tang, Yuanyuan Shi, Yiqun Sun, Anthony Kum Hoe Tung 唐一轩，施沅沅，孙亦群，Anthony Kum Hoe TungComments: Accepted by EMNLP 2025 评论：已被 EMNLP 2025 接收Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；信息检索 (cs.IR)

• [93] arXiv:2508.19740 [pdf, html, other]

  Spotlight Attention: Towards Efficient LLM Generation via Non-linear Hashing-based KV Cache Retrieval 聚焦注意力：通过基于非线性哈希的 KV 缓存检索实现高效的 LLM 生成Wenhao Li, Yuxin Zhang, Gen Luo, Haiyuan Wan, Ziyang Gong, Fei Chao, Rongrong Ji 李文昊，张宇昕，罗根，万海元，龚子阳，晁飞，纪荣荣Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [94] arXiv:2508.19724 [pdf, html, other]

  NLKI: A lightweight Natural Language Knowledge Integration Framework for Improving Small VLMs in Commonsense VQA Tasks NLKI：一种用于在常识视觉问答任务中提升小型视觉语言模型的轻量级自然语言知识整合框架Aritra Dutta, Swapnanil Mukherjee, Deepanway Ghosal, Somak AdityaSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [95] arXiv:2508.19721 [pdf, html, other]

  CAMÕES: A Comprehensive Automatic Speech Recognition Benchmark for European Portuguese CAMÕES：面向欧洲葡萄牙语的综合自动语音识别基准Carlos Carvalho, Francisco Teixeira, Catarina Botelho, Anna Pompili, Rubén Solera-Ureña, Sérgio Paulo, Mariana Julião, Thomas Rolland, John Mendonça, Diogo Pereira, Isabel Trancoso, Alberto AbadComments: Accepted to ASRU 2025 评论：已被 ASRU 2025 接收Subjects: Computation and Language (cs.CL); Audio and Speech Processing (eess.AS) 主题：计算与语言（cs.CL）；音频与语音处理（eess.AS）

• [96] arXiv:2508.19720 [pdf, html, other]

  Continuously Steering LLMs Sensitivity to Contextual Knowledge with Proxy Models 持续使用代理模型调控 LLMs 对上下文知识的敏感性Yilin Wang, Heng Wang, Yuyang Bai, Minnan Luo 王毅霖，王恒，白雨阳，罗闽南Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [97] arXiv:2508.19689 [pdf, html, other]

  Building Task Bots with Self-learning for Enhanced Adaptability, Extensibility, and Factuality 构建具自学习能力的任务机器人以增强适应性、可扩展性和事实性Xiaoying Zhang 张晓英Comments: 179 pages 注释：179 页Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [98] arXiv:2508.19667 [pdf, html, other]

  Survey of Specialized Large Language Model 专门化大型语言模型综述Chenghan Yang, Ruiyu Zhao, Yang Liu, Ling Jiang 杨承翰，赵瑞宇，刘洋，蒋岭Comments: 9 pages, 1 figures 注释：9 页，1 图Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [99] arXiv:2508.19665 [pdf, html, other]

  Automatic integration of SystemC in the FMI standard for Software-defined Vehicle design 将 SystemC 自动集成到面向软件定义车辆设计的 FMI 标准中Giovanni Pollo, Andrei Mihai Albu, Alessio Burrello, Daniele Jahier Pagliari, Cristian Tesconi, Loris Panaro, Dario Soldi, Fabio Autieri, Sara Vinco Giovanni Pollo，Andrei Mihai Albu，Alessio Burrello，Daniele Jahier Pagliari，Cristian Tesconi，Loris Panaro，Dario Soldi，Fabio Autieri，Sara VincoSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [100] arXiv:2508.19633 [pdf, html, other] [100] arXiv:2508.19633 [ pdf，html，other]

  A Symbolic Adversarial Learning Framework for Evolving Fake News Generation and Detection 一种用于生成与检测演化假新闻的符号对抗学习框架Chong Tian, Qirong Ho, Xiuying Chen 崇天、何启荣、陈秀英Comments: Accepted to EMNLP 2025 Main Conference 注释：已被接收为 EMNLP 2025 主会会议论文Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [101] arXiv:2508.19614 [pdf, html, other]

  LFD: Layer Fused Decoding to Exploit External Knowledge in Retrieval-Augmented Generation LFD：在检索增强生成中利用外部知识的层融合解码Yang Sun, Lixin Zou, Dan Luo, Zhiyong Xie, Long Zhang, Liming Dong, Yunwei Zhao, Xixun Lin, Yanxiong Lu, Chenliang Li 杨孙，邹立鑫，罗丹，谢志勇，张龙，董黎明，赵云威，林锡勋，卢艳雄，李晨亮Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [102] arXiv:2508.19594 [pdf, html, other]

  Understanding and Leveraging the Expert Specialization of Context Faithfulness in Mixture-of-Experts LLMs 理解与利用混合专家 LLMs 中语境忠实性的专家化特长Jun Bai, Minghao Tong, Yang Liu, Zixia Jia, Zilong Zheng 白俊、佟明浩、刘阳、贾子霞、郑子龙Comments: Accepted by EMNLP 2025 Main 评注：被 EMNLP 2025 主会场接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [103] arXiv:2508.19587 [pdf, html, other]

  Towards stable AI systems for Evaluating Arabic Pronunciations 迈向用于评估阿拉伯语发音的稳定人工智能系统Hadi Zaatiti, Hatem Hajri, Osama Abdullah, Nader Masmoudi 哈迪·扎蒂蒂，哈特姆·哈吉里，奥萨马·阿卜杜拉，纳迪尔·马斯穆迪Journal-ref: 4th International Conference on NLP and Machine Learning Trends 2025 期刊参考：2025 年第四届自然语言处理与机器学习趋势国际会议Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [104] arXiv:2508.19580 [pdf, html, other]

  ArgCMV: An Argument Summarization Benchmark for the LLM-era ArgCMV：一个面向 LLM 时代的论点摘要基准Omkar Gurjar, Agam Goyal, Eshwar Chandrasekharan Omkar Gurjar、Agam Goyal、Eshwar ChandrasekharanSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [105] arXiv:2508.19578 [pdf, html, other]

  Towards a Holistic and Automated Evaluation Framework for Multi-Level Comprehension of LLMs in Book-Length Contexts 面向书籍篇幅上下文中对 LLMs 多层次理解的整体化与自动化评估框架Jiaqi Deng, Yuho Lee, Nicole Hee-Yeon Kim, Hyangsuk Min, Taewon Yun, Minjeong Ban, Kim Yul, Hwanjun Song 邓佳祺、李宥豪、金希妍、闵香淑、尹太元、班敏贞、金律、宋焕浚Comments: Accepted to EMNLP 2025 (Main) 评论：已被 EMNLP 2025（主会）录用Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [106] arXiv:2508.19546 [pdf, html, other]

  Language Models Identify Ambiguities and Exploit Loopholes 语言模型识别歧义并利用漏洞Jio Choi, Mohit Bansal, Elias Stengel-EskinComments: EMNLP 2025 camera-ready; Code: this https URL 评注：EMNLP 2025 定稿；代码：此 https URLSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [107] arXiv:2508.19533 [pdf, html, other]

  Emotion Transfer with Enhanced Prototype for Unseen Emotion Recognition in Conversation 用于对话中未见情感识别的增强原型情感迁移Kun Peng, Cong Cao, Hao Peng, Guanlin Wu, Zhifeng Hao, Lei Jiang, Yanbing Liu, Philip S. Yu 彭昆，曹聪，彭浩，吴冠林，郝志峰，姜磊，刘艳兵，Philip S. YuComments: Accepted at EMNLP2025 评论：已被 EMNLP2025 接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [108] arXiv:2508.19532 [pdf, html, other]

  Alignment with Fill-In-the-Middle for Enhancing Code Generation 通过填空中间对齐以增强代码生成Houxing Ren, Zimu Lu, Weikang Shi, Haotian Hou, Yunqiao Yang, Ke Wang, Aojun Zhou, Junting Pan, Mingjie Zhan, Hongsheng Li 任厚兴, 陆梓慕, 石伟康, 侯昊天, 杨云乔, 王珂, 周奥俊, 潘俊廷, 詹明杰, 李宏胜Comments: Accepted to EMNLP 2025 (main conference) 评注：已被接受至 EMNLP 2025（主会议）Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [109] arXiv:2508.19529 [pdf, html, other]

  Blockwise SFT for Diffusion Language Models: Reconciling Bidirectional Attention and Autoregressive Decoding 面向扩散语言模型的分块微调：调和双向注意力与自回归解码Bowen Sun, Yujun Cai, Ming-Hsuan Yang, Yiwei Wang Bowen Sun、Yujun Cai、Ming-Hsuan Yang、Yiwei WangSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [110] arXiv:2508.19484 [pdf, other]

  Rule Synergy Analysis using LLMs: State of the Art and Implications 使用 LLMs 的规则协同分析：现状与影响Bahar Bateni, Benjamin Pratt, Jim WhiteheadComments: Submitted for publication at the IEEE Transactions on Games 2024, Special Issue on Large Language Models and Games (10 pages excluding appendix, 3 figures) 备注：已提交至 IEEE Transactions on Games 2024，关于大型语言模型与游戏的特刊（不含附录为 10 页，3 张图）Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [111] arXiv:2508.19481 [pdf, html, other]

  Improving Low-Resource Translation with Dictionary-Guided Fine-Tuning and RL: A Spanish-to-Wayuunaiki Study 使用词典引导的微调与强化学习改进低资源翻译：一项西班牙语到瓦尤纳伊基语的研究Manuel Mosquera, Melissa Robles, Johan Rodriguez, Ruben ManriqueSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [112] arXiv:2508.19475 [pdf, html, other]

  Automatic Question & Answer Generation Using Generative Large Language Model (LLM) 使用生成式大型语言模型 (LLM) 的自动问答生成Md. Alvee Ehsan, A.S.M Mehedi Hasan, Kefaya Benta Shahnoor, Syeda Sumaiya Tasneem Md. Alvee Ehsan、A.S.M Mehedi Hasan、Kefaya Benta Shahnoor、Syeda Sumaiya TasneemSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [113] arXiv:2508.19467 [pdf, html, other]

  Inference Gap in Domain Expertise and Machine Intelligence in Named Entity Recognition: Creation of and Insights from a Substance Use-related Dataset 命名实体识别中领域专业知识与机器智能的推理差距：与物质使用相关数据集的构建与洞见Sumon Kanti Dey, Jeanne M. Powell, Azra Ismail, Jeanmarie Perrone, Abeed Sarker Sumon Kanti Dey、Jeanne M. Powell、Azra Ismail、Jeanmarie Perrone、Abeed SarkerComments: Dataset and code: this https URL 注释：数据集和代码：此 https URLSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；信息检索 (cs.IR)

• [114] arXiv:2508.19464 [pdf, html, other]

  Bridging Language Gaps: Enhancing Few-Shot Language Adaptation 弥合语言鸿沟：增强少样本语言适应Philipp Borchert, Jochen De Weerdt, Marie-Francine MoensComments: 17 pages 注释：17 页Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [115] arXiv:2508.19428 [pdf, html, other]

  Heterogeneous LLM Methods for Ontology Learning (Few-Shot Prompting, Ensemble Typing, and Attention-Based Taxonomies) 面向本体学习的异构 LLM 方法（少样本提示、集成类型判定与基于注意力的分类法）Aleksandra Beliaeva, Temurbek RahmatullaevSubjects: Computation and Language (cs.CL); Logic in Computer Science (cs.LO); Symbolic Computation (cs.SC) 学科：计算与语言 (cs.CL); 计算机科学中的逻辑 (cs.LO); 符号计算 (cs.SC)

• [116] arXiv:2508.19427 [pdf, html, other]

  A perishable ability? The future of writing in the face of generative artificial intelligence 一种会消逝的能力？生成式人工智能面前的写作未来Evandro L. T. P. Cunha 埃万德罗·L. T. P. 庞哈Comments: 10 pages 备注：10 页Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Human-Computer Interaction (cs.HC) 学科：计算与语言（cs.CL）；人工智能（cs.AI）；计算机与社会（cs.CY）；人机交互（cs.HC）

• [117] arXiv:2508.19402 [pdf, html, other]

  One Joke to Rule them All? On the (Im)possibility of Generalizing Humor 一笑打尽？关于普遍化幽默的（不）可能性Mor Turgeman, Chen Shani, Dafna Shahaf Mor Turgeman、Chen Shani、Dafna ShahafSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [118] arXiv:2508.19372 [pdf, html, other]

  Database Entity Recognition with Data Augmentation and Deep Learning 数据库实体识别：数据增强与深度学习Zikun Fu, Chen Yang, Kourosh Davoudi, Ken Q. PuComments: 6 pages, 5 figures. Accepted at IEEE 26th International Conference on Information Reuse and Integration for Data Science (IRI 2025), San Jose, California, August 6-8, 2025 备注：6 页，5 幅图。已被 IEEE 第 26 届国际信息重用与集成数据科学会议（IRI 2025）接收，会议地点：加利福尼亚圣何塞，时间：2025 年 8 月 6-8 日Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Databases (cs.DB); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；数据库（cs.DB）；机器学习（cs.LG）

• [119] arXiv:2508.19363 [pdf, html, other]

  LongReasonArena: A Long Reasoning Benchmark for Large Language Models LongReasonArena：面向大型语言模型的长推理基准Jiayu Ding, Shuming Ma, Lei Cui, Nanning Zheng, Furu Wei 丁佳钰，马澍明，崔雷，郑南宁，魏拂Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [120] arXiv:2508.19359 [pdf, html, other]

  Reflective Agreement: Combining Self-Mixture of Agents with a Sequence Tagger for Robust Event Extraction 反思式一致性：将代理自混合与序列标注器相结合以实现稳健的事件抽取Fatemeh Haji, Mazal Bethany, Cho-Yu Jason Chiang, Anthony Rios, Peyman Najafirad 法蒂玛·哈吉，玛扎尔·贝萨尼，张卓羽（Cho-Yu Jason Chiang），安东尼·里奥斯，佩伊曼·纳贾菲拉德Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [121] arXiv:2508.19357 [pdf, html, other]

  Context-Adaptive Synthesis and Compression for Enhanced Retrieval-Augmented Generation in Complex Domains 面向复杂领域的改进检索增强生成的上下文自适应合成与压缩Peiran Zhou, Junnan Zhu, Yichen Shen, Ruoxi Yu 周沛然, 朱俊南, 沈奕晨, 余若曦Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [122] arXiv:2508.19282 [pdf, html, other]

  CORE: Lossless Compression for Retrieval-Augmented LLMs via Reinforcement Learning CORE：通过强化学习实现面向检索增强型 LLMs 的无损压缩Ziqiang Cui, Yunpeng Weng, Xing Tang, Peiyang Liu, Shiwei Li, Bowei He, Jiamin Chen, Xiuqiang He, Chen Ma 崔子强, 翁云鹏, 唐星, 刘培阳, 李世炜, 何博伟, 陈佳敏, 何修强, 马晨Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [123] arXiv:2508.19279 [pdf, other]

  FLAIRR-TS – Forecasting LLM-Agents with Iterative Refinement and Retrieval for Time Series FLAIRR-TS —— 用于时间序列的迭代精炼与检索的 LLM-Agent 预测方法Gunjan Jalori, Preetika Verma, Sercan Ö ArıkComments: EMNLP 备注：EMNLPSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [124] arXiv:2508.19274 [pdf, html, other]

  Leveraging Language Models and Machine Learning in Verbal Autopsy Analysis 在口述死亡调查分析中利用语言模型和机器学习Yue Chu 楚悦Comments: Ph.D. dissertation submitted to The Ohio State University, August 2025 注释：博士论文，提交给俄亥俄州立大学，2025 年 8 月Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [125] arXiv:2508.19272 [pdf, html, other]

  RAGAPHENE: A RAG Annotation Platform with Human Enhancements and Edits RAGAPHENE：一个具有人工增强与编辑功能的 RAG 注释平台Kshitij Fadnis, Sara Rosenthal, Maeda Hanafi, Yannis Katsis, Marina DanilevskySubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [126] arXiv:2508.19271 [pdf, html, other]

  Rethinking Reasoning in LLMs: Neuro-Symbolic Local RetoMaton Beyond ICL and CoT 重新思考 LLMs 中的推理：超越 ICL 和 CoT 的神经符号本地 RetoMatonRushitha Santhoshi Mamidala, Anshuman Chhabra, Ankur Mali Rushitha Santhoshi Mamidala，Anshuman Chhabra，Ankur MaliSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [127] arXiv:2508.19270 [pdf, html, other] [127] arXiv:2508.19270 [ pdf，html，other]

  Whisper based Cross-Lingual Phoneme Recognition between Vietnamese and English 基于 Whisper 的越南语与英语之间的跨语言音素识别Nguyen Huu Nhat Minh, Tran Nguyen Anh, Truong Dinh Dung, Vo Van Nam, Le Pham TuyenSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [128] arXiv:2508.19268 [pdf, html, other]

  MultiPL-MoE: Multi-Programming-Lingual Extension of Large Language Models through Hybrid Mixture-of-Experts MultiPL-MoE：通过混合专家模型扩展大型语言模型的多编程语言能力Qing Wang, Xue Han, Jiahui Wang, Lehao Xing, Qian Hu, Lianlian Zhang, Chao Deng, Junlan FengSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [129] arXiv:2508.20083 (cross-list from cs.CR) [pdf, other] [129] arXiv:2508.20083（从 cs.CR 交叉列出）[ pdf，其他]

  Disabling Self-Correction in Retrieval-Augmented Generation via Stealthy Retriever Poisoning 通过隐蔽的检索器投毒在检索增强生成中禁用自我纠正Yanbo Dai, Zhenlan Ji, Zongjie Li, Kuan Li, Shuai Wang 戴彦博，季振兰，李宗杰，李寛，王帅Subjects: Cryptography and Security (cs.CR); Computation and Language (cs.CL) 主题：密码学与安全（cs.CR）；计算与语言（cs.CL）

• [130] arXiv:2508.20032 (cross-list from cs.LG) [pdf, html, other] [130] arXiv:2508.20032（跨列表自 cs.LG）[ pdf, html, other]

  Pruning Strategies for Backdoor Defense in LLMs 针对 LLMs 后门防御的剪枝策略Santosh Chapagain, Shah Muhammad Hamdi, Soukaina Filali Boubrahimi Santosh Chapagain、Shah Muhammad Hamdi、Soukaina Filali BoubrahimiComments: Accepted in CIKM ‘25: The 34th ACM International Conference on Information and Knowledge Management Proceedings 备注：已被 CIKM ‘25 接收：第 34 届 ACM 信息与知识管理国际会议论文集Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL) 学科：机器学习（cs.LG）；计算与语言（cs.CL）

• [131] arXiv:2508.20019 (cross-list from cs.LG) [pdf, html, other] [131] arXiv:2508.20019（从 cs.LG 交叉列出）[ pdf, html, other]

  Symphony: A Decentralized Multi-Agent Framework for Scalable Collective Intelligence 交响曲：一个用于可扩展集体智能的去中心化多智能体框架Ji Wang, Kashing Chen, Xinyuan Song, Ke Zhang, Lynn Ai, Eric Yang, Bill Shi 王骥，陈卡星，宋新远，张科，艾琳，杨逸，史比尔Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multiagent Systems (cs.MA) 主题：机器学习 (cs.LG)；人工智能 (cs.AI)；计算与语言 (cs.CL)；多智能体系统 (cs.MA)

• [132] arXiv:2508.20018 (cross-list from cs.AI) [pdf, html, other] [132] arXiv:2508.20018（从 cs.AI 交叉列出）[ pdf, html, other]

  SWIRL: A Staged Workflow for Interleaved Reinforcement Learning in Mobile GUI Control SWIRL：一种用于移动 GUI 控制的交错强化学习分阶段工作流Quanfeng Lu, Zhantao Ma, Shuai Zhong, Jin Wang, Dahai Yu, Michael K. Ng, Ping Luo 卢全峰，马展涛，钟帅，王劲，俞大海，伍诗博（Michael K. Ng），骆平Comments: 28 pages, 12 figures 评论：28 页，12 幅图Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Multiagent Systems (cs.MA) 主题：人工智能 (cs.AI)；计算与语言 (cs.CL)；计算机视觉与模式识别 (cs.CV)；多智能体系统 (cs.MA)

• [133] arXiv:2508.19999 (cross-list from cs.LG) [pdf, html, other] [133] arXiv:2508.19999（从 cs.LG 交叉列出）[ pdf，html，其他]

  Linear-Time Demonstration Selection for In-Context Learning via Gradient Estimation 通过梯度估计进行上下文学习的线性时间示例选择Ziniu Zhang, Zhenshuo Zhang, Dongyue Li, Lu Wang, Jennifer Dy, Hongyang R. Zhang 张子牛, 张振硕, 李东岳, 王璐, Jennifer Dy, 张洪洋Comments: 19 pages. To appear in EMNLP'25 评论：19 页。将发表于 EMNLP'25Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：机器学习 (cs.LG); 人工智能 (cs.AI); 计算与语言 (cs.CL)

• [134] arXiv:2508.19990 (cross-list from cs.LG) [pdf, html, other] [134] arXiv:2508.19990（跨列表自 cs.LG）[ pdf, html, other ]

  Self-Supervised Pre-Training with Equilibrium Constraints 带有平衡约束的自监督预训练Xiaodong Cui, A F M Saif, Brian Kingsbury, Tianyi Chen 崔晓东、A F M Saif、布赖恩·金斯伯里、陈天毅Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL) 学科：机器学习（cs.LG）；计算与语言（cs.CL）

• [135] arXiv:2508.19972 (cross-list from cs.CV) [pdf, html, other] [135] arXiv:2508.19972（来自 cs.CV 的交叉列表）[ pdf，html，其他]

  GLSim: Detecting Object Hallucinations in LVLMs via Global-Local Similarity GLSim：通过全局-局部相似性检测大型视觉语言模型中的物体幻觉Seongheon Park, Yixuan Li 朴成炫，李奕轩Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：计算机视觉与模式识别（cs.CV）；人工智能（cs.AI）；计算与语言（cs.CL）

• [136] arXiv:2508.19944 (cross-list from cs.CV) [pdf, html, other] [136] arXiv:2508.19944（跨列自 cs.CV）[ pdf，html，other]

  KRETA: A Benchmark for Korean Reading and Reasoning in Text-Rich VQA Attuned to Diverse Visual Contexts KRETA：面向多样视觉语境的文本丰富型视觉问答中韩语阅读与推理基准Taebaek Hwang, Minseo Kim, Gisang Lee, Seonuk Kim, Hyunjun Eun 黄泰栢，金敏瑞，李基桑，金先旭，殷炫俊Subjects: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL) 主题：计算机视觉与模式识别 (cs.CV)；计算与语言 (cs.CL)

• [137] arXiv:2508.19843 (cross-list from cs.CR) [pdf, html, other] [137] arXiv:2508.19843（从 cs.CR 交叉列出）[ pdf，html，其他]

  SoK: Large Language Model Copyright Auditing via Fingerprinting SoK：通过指纹技术对大型语言模型进行版权审计Shuo Shao, Yiming Li, Yu He, Hongwei Yao, Wenyuan Yang, Dacheng Tao, Zhan Qin 邵硕，李亦鸣，何昱，姚宏伟，杨文远，陶大程，秦展Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：密码学与安全（cs.CR）；人工智能（cs.AI）；计算与语言（cs.CL）

• [138] arXiv:2508.19827 (cross-list from cs.AI) [pdf, html, other] [138] arXiv:2508.19827（从 cs.AI 交叉列出）[ pdf，html，other]

  Analysing Chain of Thought Dynamics: Active Guidance or Unfaithful Post-hoc Rationalisation? 分析思路链动态：主动引导还是不真实的事后合理化？Samuel Lewis-Lim, Xingwei Tan, Zhixue Zhao, Nikolaos Aletras Samuel Lewis-Lim、Xingwei Tan、Zhixue Zhao、Nikolaos AletrasComments: Accepted at EMNLP 2025 Main Conference 评论：已被 EMNLP 2025 主会议接受Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [139] arXiv:2508.19697 (cross-list from cs.CR) [pdf, html, other] [139] arXiv:2508.19697（从 cs.CR 交叉列出）[ pdf，html，other]

  Safety Alignment Should Be Made More Than Just A Few Attention Heads 安全对齐不应仅限于少数注意力头Chao Huang, Zefeng Zhang, Juewei Yue, Quangang Li, Chuang Zhang, Tingwen Liu 黄超，张泽锋，岳珏玮，李全刚，张闯，刘廷文Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：密码学与安全（cs.CR）；人工智能（cs.AI）；计算与语言（cs.CL）

• [140] arXiv:2508.19619 (cross-list from math.CO) [pdf, html, other] [140] arXiv:2508.19619（从 math.CO 交叉列出）[ pdf，html，其他]

  Word Chain Generators for Prefix Normal Words 前缀正规词的单词链生成器Duncan Adamson, Moritz Dudey, Pamela Fleischmann, Annika HuchSubjects: Combinatorics (math.CO); Computation and Language (cs.CL) 学科：组合数学（math.CO）；计算与语言（cs.CL）

• [141] arXiv:2508.19611 (cross-list from cs.AI) [pdf, other] [141] arXiv:2508.19611（从 cs.AI 交叉列出）[ pdf，other]

  Instructional Agents: LLM Agents on Automated Course Material Generation for Teaching Faculties 教学代理：用于教学教师的自动化课程材料生成的 LLM 代理Huaiyuan Yao, Wanpeng Xu, Justin Turnau, Nadia Kellam, Hua Wei 姚怀远，许万鹏，Justin Turnau，Nadia Kellam，魏华Comments: 18 pages, 9 figures 注释：18 页，9 幅图Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [142] arXiv:2508.19558 (cross-list from cs.SE) [pdf, html, other] [142] arXiv:2508.19558（从 cs.SE 交叉列出）[ pdf, html, other]

  Functional Consistency of LLM Code Embeddings: A Self-Evolving Data Synthesis Framework for Benchmarking LLM 代码嵌入的功能一致性：用于基准测试的自我演化数据合成框架Zhuohao Li, Wenqing Chen, Jianxing Yu, Zhichao Lu 李卓昊，陈文清，郁建兴，卢志超Subjects: Software Engineering (cs.SE); Computation and Language (cs.CL); Programming Languages (cs.PL) 学科：软件工程 (cs.SE); 计算与语言 (cs.CL); 编程语言 (cs.PL)

• [143] arXiv:2508.19492 (cross-list from cs.CY) [pdf, html, other] [143] arXiv:2508.19492（从 cs.CY 交叉列出）[ pdf, html, other]

  Geopolitical Parallax: Beyond Walter Lippmann Just After Large Language Models 地缘政治视差：在大型语言模型之后，超越沃尔特·李普曼Mehmet Can Yavuz, Humza Gohar Kabir, Aylin ÖzkanComments: 7 pages, 4 figures, 7 tables 注释：7 页，4 幅图，7 张表Subjects: Computers and Society (cs.CY); Computation and Language (cs.CL) 学科：计算机与社会（cs.CY）；计算与语言（cs.CL）

• [144] arXiv:2508.19321 (cross-list from cs.CR) [pdf, html, other] [144] arXiv:2508.19321（从 cs.CR 交叉列出）[ pdf, html, 其他]

  An Investigation on Group Query Hallucination Attacks 关于群体查询幻觉攻击的研究Kehao Miao, Xiaolong JinSubjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：密码学与安全（cs.CR）；人工智能（cs.AI）；计算与语言（cs.CL）

• [145] arXiv:2508.19316 (cross-list from cs.AI) [pdf, html, other] [145] arXiv:2508.19316（来自 cs.AI 的交叉列表）[ pdf, html, other]

  Sycophancy as compositions of Atomic Psychometric Traits 谄媚行为作为原子心理测量特质的组合Shreyans Jain, Alexandra Yost, Amirali AbdullahComments: 8 pages, 4 figures 评论：8 页，4 幅图Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：人工智能 (cs.AI)；计算与语言 (cs.CL)；机器学习 (cs.LG)

• [146] arXiv:2508.19294 (cross-list from cs.CV) [pdf, html, other] [146] arXiv:2508.19294（从 cs.CV 交叉列出）[ pdf，html，other]

  Object Detection with Multimodal Large Vision-Language Models: An In-depth Review 多模态大视觉-语言模型的目标检测：一项深入综述Ranjan Sapkota, Manoj Karkee Ranjan Sapkota，Manoj KarkeeComments: First Peer Reviewed Review Paper for Object Detection with Vision-Language Models (VLMs) 评注：面向目标检测的视-语言模型（VLMs）的首篇同行评审综述论文Journal-ref: Information Fusion, 2025 期刊参考：Information Fusion，2025Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：计算机视觉与模式识别（cs.CV）；人工智能（cs.AI）；计算与语言（cs.CL）

• [147] arXiv:2508.19269 (cross-list from cs.CY) [pdf, html, other] [147] arXiv:2508.19269（从 cs.CY 交叉列出）[ pdf，html，other]

  Should LLMs be WEIRD? Exploring WEIRDness and Human Rights in Large Language Models LLMs 应该是 WEIRD 吗？在大型语言模型中探讨 WEIRD 性与人权Ke Zhou, Marios Constantinides, Daniele Quercia 周科, Marios Constantinides, Daniele QuerciaComments: This paper has been accepted in AIES 2025 评论：本文已被 AIES 2025 接收Subjects: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：计算机与社会（cs.CY）；人工智能（cs.AI）；计算与语言（cs.CL）

• [148] arXiv:2508.19262 (cross-list from cs.SD) [pdf, html, other] [148] arXiv:2508.19262（从 cs.SD 交叉列出）[ pdf, html, other]

  Beat-Based Rhythm Quantization of MIDI Performances 基于节拍的 MIDI 演奏节奏量化Maximilian Wachter, Sebastian Murgul, Michael Heizmann Maximilian Wachter、Sebastian Murgul、Michael HeizmannComments: Accepted to the Late Breaking Demo Papers of the 1st AES International Conference on Artificial Intelligence and Machine Learning for Audio (AIMLA LBDP), 2025 备注：已被接受为首届 AES 人工智能与机器学习用于音频国际会议（AIMLA）晚期突破演示论文（LBDP），2025 年Subjects: Sound (cs.SD); Computation and Language (cs.CL); Multimedia (cs.MM); Audio and Speech Processing (eess.AS) 主题：声音 (cs.SD)；计算与语言 (cs.CL)；多媒体 (cs.MM)；音频与语音处理 (eess.AS)

• [149] arXiv:2508.19259 (cross-list from cs.HC) [pdf, other] [149] arXiv:2508.19259（从 cs.HC 交叉列出）[pdf，其他]

  Capabilities of GPT-5 across critical domains: Is it the next breakthrough? GPT-5 在关键领域的能力：它是下一个突破吗？Georgios P. GeorgiouSubjects: Human-Computer Interaction (cs.HC); Computation and Language (cs.CL) 主题：人机交互（cs.HC）；计算与语言（cs.CL）

• [150] arXiv:2508.19227 [pdf, other] [150] arXiv:2508.19227 [ pdf，其他]

  Generative Interfaces for Language Models 面向生成式语言模型的界面Jiaqi Chen, Yanzhe Zhang, Yutong Zhang, Yijia Shao, Diyi Yang 陈佳祺、张燕哲、张雨桐、邵一佳、杨棣怡Comments: Preprint 注释：预印本Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；人机交互 (cs.HC)

• [151] arXiv:2508.19221 [pdf, html, other]

  Evaluating the Evaluators: Are readability metrics good measures of readability? 评估评估者：可读性指标是衡量可读性的良好方法吗？Isabel Cachola, Daniel Khashabi, Mark Dredze 伊莎贝尔·卡乔拉、丹尼尔·卡沙比、马克·德雷泽Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [152] arXiv:2508.19205 [pdf, html, other]

  VibeVoice Technical Report VibeVoice 技术报告Zhiliang Peng, Jianwei Yu, Wenhui Wang, Yaoyao Chang, Yutao Sun, Li Dong, Yi Zhu, Weijiang Xu, Hangbo Bao, Zehua Wang, Shaohan Huang, Yan Xia, Furu Wei 彭志良，于建伟，王文辉，常瑶瑶，孙禹滔，董丽，祝怡，徐伟江，鲍航博，王泽华，黄少翰，夏岩，魏福如Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Sound (cs.SD); Audio and Speech Processing (eess.AS) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；声音（cs.SD）；音频与语音处理（eess.AS）

• [153] arXiv:2508.19202 [pdf, html, other]

  Demystifying Scientific Problem-Solving in LLMs by Probing Knowledge and Reasoning 揭开通过探测知识与推理来理解 LLMs 的科学问题解决之谜Alan Li, Yixin Liu, Arpan Sarkar, Doug Downey, Arman CohanComments: 28 pages, 16 figures 注释：28 页，16 幅图Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [154] arXiv:2508.19111 [pdf, html, other]

  Do LVLMs Know What They Know? A Systematic Study of Knowledge Boundary Perception in LVLMs 大型视觉语言模型（LVLMs）知道它们所知道的吗？关于 LVLMs 知识边界感知的系统性研究Zhikai Ding, Shiyu Ni, Keping Bi 丁志凯，倪世宇，毕克平Comments: EMNLP2025 Findings 注释：EMNLP2025 FindingsSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [155] arXiv:2508.19099 [pdf, html, other]

  Beyond the Black Box: Integrating Lexical and Semantic Methods in Quantitative Discourse Analysis with BERTopic 超越黑箱：在定量话语分析中结合词汇和语义方法的 BERTopic 应用Thomas Compton 托马斯·康普顿Comments: 5 pages conference paper, 4 tables 评论：5 页会议论文，4 张表Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [156] arXiv:2508.19093 [pdf, other]

  Retrieval-Augmented Generation for Natural Language Art Provenance Searches in the Getty Provenance Index 用于盖蒂来源索引中自然语言艺术来源追溯搜索的检索增强生成Mathew HenricksonSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [157] arXiv:2508.19089 [pdf, html, other]

  It’s All About In-Context Learning! Teaching Extremely Low-Resource Languages to LLMs 一切都关乎上下文学习！教会 LLMs 极低资源语言Yue Li, Zhixue Zhao, Carolina Scarton 李悦，赵志学，Carolina ScartonComments: Accepted by EMNLP 2025 评论：已被 EMNLP 2025 接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [158] arXiv:2508.19077 [pdf, html, other]

  “Where does it hurt?” – Dataset and Study on Physician Intent Trajectories in Doctor Patient Dialogues “哪里痛？”——医生与病人对话中医生意图轨迹的数据集与研究Tom Röhr, Soumyadeep Roy, Fares Al Mohamad, Jens-Michalis Papaioannou, Wolfgang Nejdl, Felix Gers, Alexander LöserComments: Accepted at ECAI 2025 评注：被 ECAI 2025 接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [159] arXiv:2508.19076 [pdf, html, other]

  HiPlan: Hierarchical Planning for LLM-Based Agents with Adaptive Global-Local Guidance HiPlan：面向基于 LLM 代理的分层规划及自适应全局-局部引导Ziyue Li, Yuan Chang, Gaihong Yu, Xiaoqiu Le 李子悦，常远，喻盖鸿，乐晓秋Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [160] arXiv:2508.19026 [pdf, html, other]

  MovieCORE: COgnitive REasoning in Movies MovieCORE：电影中的认知推理Gueter Josmy Faure, Min-Hung Chen, Jia-Fong Yeh, Ying Cheng, Hung-Ting Su, Yung-Hao Tang, Shang-Hong Lai, Winston H. Hsu Gueter Josmy Faure、Min-Hung Chen、Jia-Fong Yeh、Ying Cheng、Hung-Ting Su、Yung-Hao Tang、Shang-Hong Lai、Winston H. HsuComments: Accepted for EMNLP'2025 Main Conference. Project Page: this https URL 评论：已被接收为 EMNLP 2025 主会会议论文。项目页面：此 https URLSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；计算机视觉与模式识别 (cs.CV)

• [161] arXiv:2508.18992 [pdf, html, other]

  Automatic Prompt Optimization with Prompt Distillation 通过提示蒸馏的自动提示优化Viktor N. Zhuravlev, Artur R. Khairullin, Ernest A. Dyagin, Alena N. Sitkina, Nikita I. Kulin Viktor N. Zhuravlev、Artur R. Khairullin、Ernest A. Dyagin、Alena N. Sitkina、Nikita I. KulinSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [162] arXiv:2508.18988 [pdf, html, other]

  Interpretable by AI Mother Tongue: Native Symbolic Reasoning in Neural Models 可被人工智能以母语解释：神经模型中的本土符号推理Hung Ming Liu 刘鸿明Comments: 25 pages, 9 figures. The AI Intuition Explorer dashboard is available at: this https URL 注释：25 页，9 张图。AI Intuition Explorer 仪表板可通过此 https URL 访问。Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [163] arXiv:2508.18929 [pdf, html, other]

  Diverse And Private Synthetic Datasets Generation for RAG evaluation: A multi-agent framework 用于 RAG 评估的多样且私密的合成数据集生成：一个多代理框架Ilias Driouich, Hongliu Cao, Eoin Thomas Ilias Driouich、Hongliu Cao、Eoin ThomasComments: ECAI 2025 TRUST AI workshop 评论：ECAI 2025 TRUST AI 研讨会Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [164] arXiv:2508.18916 [pdf, html, other]

  Affective Polarization across European Parliaments 情感极化在欧洲议会中的表现Bojan Evkoski, Igor Mozetič, Nikola Ljubešić, Petra Kralj Novak Bojan Evkoski、Igor Mozetič、Nikola Ljubešić、Petra Kralj NovakComments: 6 pages, 4 figures 评论：6 页，4 幅图Subjects: Computation and Language (cs.CL); Social and Information Networks (cs.SI) 学科：计算与语言（cs.CL）；社会与信息网络（cs.SI）

• [165] arXiv:2508.18872 [pdf, html, other]

  Empowering Computing Education Researchers Through LLM-Assisted Content Analysis 通过 LLM 辅助的内容分析赋能计算教育研究者Laurie Gale, Sebastian Mateos Nicolajsen 劳里·盖尔, 塞巴斯蒂安·马特奥斯·尼科拉森Comments: 7 pages, 2 figures 评论：7 页，2 张图Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [166] arXiv:2508.18870 [pdf, html, other]

  ReflectivePrompt: Reflective evolution in autoprompting algorithms ReflectivePrompt：自动提示算法中的反思性演化Viktor N. Zhuravlev, Artur R. Khairullin, Ernest A. Dyagin, Alena N. Sitkina, Nikita I. Kulin Viktor N. Zhuravlev、Artur R. Khairullin、Ernest A. Dyagin、Alena N. Sitkina、Nikita I. KulinSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [167] arXiv:2508.18847 [pdf, html, other]

  ConfTuner: Training Large Language Models to Express Their Confidence Verbally ConfTuner：训练大型语言模型以用语言表达其置信度Yibo Li, Miao Xiong, Jiaying Wu, Bryan Hooi 李逸博，熊淼，吴佳莹，胡培然（Bryan Hooi）Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [168] arXiv:2508.18824 [pdf, html, other]

  Arrows of Math Reasoning Data Synthesis for Large Language Models: Diversity, Complexity and Correctness 数学推理数据合成用于大型语言模型：多样性、复杂性与正确性Sirui Chen, Changxin Tian, Binbin Hu, Kunlong Chen, Ziqi Liu, Zhiqiang Zhang, Jun ZhouSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [169] arXiv:2508.18819 [pdf, html, other]

  LLM-based Contrastive Self-Supervised AMR Learning with Masked Graph Autoencoders for Fake News Detection 基于 LLM 的对比自监督 AMR 学习与用于假新闻检测的掩蔽图自编码器Shubham Gupta, Shraban Kumar Chatterjee, Suman KunduSubjects: Computation and Language (cs.CL); Social and Information Networks (cs.SI) 学科：计算与语言（cs.CL）；社会与信息网络（cs.SI）

• [170] arXiv:2508.18791 [pdf, html, other]

  LaTeXTrans: Structured LaTeX Translation with Multi-Agent Coordination LaTeXTrans：具有多代理协调的结构化 LaTeX 翻译Ziming Zhu, Chenglong Wang, Shunjie Xing, Yifu Huo, Fengning Tian, Quan Du, Di Yang, Chunliang Zhang, Tong Xiao, Jingbo Zhu 朱子明，王成龙，邢舜杰，霍奕甫，田凤宁，杜泉，杨迪，张春良，萧彤，朱敬博Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [171] arXiv:2508.18783 [pdf, html, other]

  Controllable Conversational Theme Detection Track at DSTC 12 DSTC 12 可控会话主题检测赛道Igor Shalyminov, Hang Su, Jake Vincent, Siffi Singh, Jason Cai, James Gung, Raphael Shu, Saab Mansour Igor Shalyminov、Hang Su、Jake Vincent、Siffi Singh、Jason Cai、James Gung、Raphael Shu、Saab MansourComments: DSTC12@SigDial2025; data and code available at this https URL 备注：DSTC12@SigDial2025；数据和代码可在此 https URL 获取Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [172] arXiv:2508.18780 [pdf, html, other]

  Harnessing Rule-Based Reinforcement Learning for Enhanced Grammatical Error Correction 利用基于规则的强化学习提升语法错误纠正Yilin Li, Xunjian Yin, Yilin Chen, Xiaojun Wan 李一琳, 尹迅健, 陈一琳, 万晓军Comments: Code will be released upon publication 备注：代码将在发表时公开Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [173] arXiv:2508.18773 [pdf, html, other]

  ThinkDial: An Open Recipe for Controlling Reasoning Effort in Large Language Models ThinkDial：一种用于控制大型语言模型推理努力的开放配方Qianyu He, Siyu Yuan, Xuefeng Li, Mingxuan Wang, Jiangjie Chen 何千宇，袁思宇，李雪峰，王明轩，陈江杰Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [174] arXiv:2508.18748 [pdf, html, other] [174] arXiv:2508.18748 [ pdf、html、其他]

  Chronological Passage Assembling in RAG framework for Temporal Question Answering 在 RAG 框架中为时序问答进行按时间顺序片段组装Byeongjeong Kim, Jeonghyun Park, Joonho Yang, Hwanhee Lee 金秉井，朴正炫，杨俊浩，李奂熙Comments: 7 pages, 3 figures 注释：7 页，3 幅图Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [175] arXiv:2508.18740 [pdf, html, other]

  M3HG: Multimodal, Multi-scale, and Multi-type Node Heterogeneous Graph for Emotion Cause Triplet Extraction in Conversations M3HG：用于对话中情感原因三元组抽取的多模态、多尺度、多类型节点异构图Qiao Liang, Ying Shen, Tiantian Chen, Lin Zhang 乔亮，沈颖，陈甜甜，张琳Comments: 16 pages, 8 figures. Accepted to Findings of ACL 2025 评论：16 页，8 幅图。被《ACL 2025 研究成果》接收Journal-ref: Findings of ACL 2025 (2025) 11416-11431 期刊参考：《ACL 2025 研究成果》（2025）11416-11431Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [176] arXiv:2508.18739 [pdf, html, other]

  Beyond Quality: Unlocking Diversity in Ad Headline Generation with Large Language Models 超越质量：用大型语言模型在广告标题生成中解锁多样性Chang Wang, Siyu Yan, Depeng Yuan, Yuqi Chen, Yanhua Huang, Yuanhang Zheng, Shuhao Li, Yinqi Zhang, Kedi Chen, Mingrui Zhu, Ruiwen Xu Chang Wang、Siyu Yan、Depeng Yuan、Yuqi Chen、Yanhua Huang、Yuanhang Zheng、Shuhao Li、Yinqi Zhang、Kedi Chen、Mingrui Zhu、Ruiwen XuSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [177] arXiv:2508.18715 [pdf, html, other]

  EMMM, Explain Me My Model! Explainable Machine Generated Text Detection in Dialogues EMMM，解释我的模型！对话中机器生成文本的可解释检测Angela Yifei Yuan, Haoyi Li, Soyeon Caren Han, Christopher Leckie Angela Yifei Yuan、Haoyi Li、Soyeon Caren Han、Christopher LeckieComments: 15 pages 注释：15 页Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [178] arXiv:2508.18709 [pdf, html, other]

  Filtering for Creativity: Adaptive Prompting for Multilingual Riddle Generation in LLMs 为创造力过滤：用于 LLMs 多语言谜语生成的自适应提示Duy Le, Kent Ziti, Evan Girard-Sun, Sean O’Brien, Vasu Sharma, Kevin ZhuSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [179] arXiv:2508.18701 [pdf, html, other]

  Attention2Probability: Attention-Driven Terminology Probability Estimation for Robust Speech-to-Text System Attention2Probability：基于注意力的术语概率估计用于鲁棒的语音到文本系统Yanfan Du, Jun Zhang, Bin Wang, Jin Qiu, Lu Huang, Yuan Ge, Xiaoqian Liu, Tong Xiao, Jingbo Zhu 杜艳帆，张军，王斌，邱进，黄璐，葛元，刘晓倩，肖通，祝景波Comments: 9 pages, 4 figures, 5 tables 备注：9 页，4 幅图，5 张表Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [180] arXiv:2508.18687 [pdf, html, other]

  Knowing or Guessing? Robust Medical Visual Question Answering via Joint Consistency and Contrastive Learning 知道还是猜测？通过联合一致性与对比学习实现稳健的医学视觉问答Songtao Jiang, Yuxi Chen, Sibo Song, Yan Zhang, Yeying Jin, Yang Feng, Jian Wu, Zuozhu Liu 宋涛 江, 陈宇曦, 宋思博, 张燕, 金叶莹, 冯阳, 吴剑, 刘作柱Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [181] arXiv:2508.18673 [pdf, html, other]

  Tailored Teaching with Balanced Difficulty: Elevating Reasoning in Multimodal Chain-of-Thought via Prompt Curriculum 定制教学与难度平衡：通过提示课程提升多模态链式思维的推理能力Xinglong Yang, Quan Feng, Zhongying Pan, Xiang Chen, Yu Tian, Wentong Li, Shuofei Qiao, Yuxia Geng, Xingyu Zhao, Sheng-Jun Huang 杨兴龙，冯泉，潘中英，陈翔，田宇，李文彤，乔硕飞，耿玉霞，赵星宇，黄胜军Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Multimedia (cs.MM) 学科：计算与语言（cs.CL）；人工智能（cs.AI）；多媒体（cs.MM）

• [182] arXiv:2508.18655 [pdf, html, other]

  Emotion Omni: Enabling Empathetic Speech Response Generation through Large Language Models Emotion Omni：通过大语言模型实现具有同理心的语音响应生成Haoyu Wang, Guangyan Zhang, Jiale Chen, Jingyu Li, Yuehai Wang, Yiwen Guo 王浩宇，张光艳，陈佳乐，李静宇，王跃海，郭奕文Comments: 5 pages, 1 figure, submitted to ICASSP 2026 注释：5 页，1 图，已提交至 ICASSP 2026Subjects: Computation and Language (cs.CL); Sound (cs.SD); Audio and Speech Processing (eess.AS) 学科：计算与语言（cs.CL）；声音（cs.SD）；音频与语音处理（eess.AS）

• [183] arXiv:2508.18651 [pdf, html, other]

  Breaking the Trade-Off Between Faithfulness and Expressiveness for Large Language Models 打破大型语言模型在忠实性与表达性之间的权衡Chenxu Yang, Qingyi Si, Zheng Lin 陈旭阳、斯庆奕、林正Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [184] arXiv:2508.18648 [pdf, html, other]

  Thinking Before You Speak: A Proactive Test-time Scaling Approach 先思考再发言：一种主动的测试时缩放方法Cong Liu, Wenchang Chai, Hejun Wu, Yan Pan, Pengxu Wei, Liang Lin 刘聪、柴文昌、吴和俊、潘岩、魏鹏旭、林亮Journal-ref: EMNLP 2025 期刊引用：EMNLP 2025Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [185] arXiv:2508.18609 [pdf, html, other]

  Scaling Laws for Task-Stratified Knowledge in Post-Training Quantized Large Language Models 后训练量化大型语言模型中任务分层知识的规模律Chenxi Zhou, Pengfei Cao, Jiang Li, Jun Zhao, Kang Liu 周晨曦，曹鹏飞，李江，赵骏，刘康Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [186] arXiv:2508.18607 [pdf, html, other]

  A New NMT Model for Translating Clinical Texts from English to Spanish 一种用于将临床文本从英语翻译为西班牙语的新神经机器翻译模型Rumeng Li, Xun Wang, Hong YuComments: This work was accepted by the Machine Learning for Health (ML4H) Workshop at NeurIPS 2018 评论：该工作已被 2018 年 NeurIPS 的机器学习与健康（ML4H）研讨会接受Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [187] arXiv:2508.18598 [pdf, html, other]

  What do language models model? Transformers, automata, and the format of thought 语言模型建模的是什么？变换器、自动机与思维的形式Colin Klein 科林·克莱因Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [188] arXiv:2508.18569 [pdf, html, other]

  The Mind’s Eye: A Multi-Faceted Reward Framework for Guiding Visual Metaphor Generation 心灵之眼：用于引导视觉隐喻生成的多面奖励框架Girish A. Koushik, Fatemeh Nazarieh, Katherine Birch, Shenbin Qian, Diptesh Kanojia Girish A. Koushik、Fatemeh Nazarieh、Katherine Birch、Shenbin Qian、Diptesh KanojiaComments: Under Review 评论：审查中Subjects: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV) 学科：计算与语言 (cs.CL)；计算机视觉与模式识别 (cs.CV)

• [189] arXiv:2508.18549 [pdf, html, other]

  COMET-poly: Machine Translation Metric Grounded in Other Candidates COMET-poly：以其他候选译文为基础的机器翻译评估指标Maike Züfle, Vilém Zouhar, Tu Anh Dinh, Felipe Maia Polo, Jan Niehues, Mrinmaya Sachan Maike Züfle、Vilém Zouhar、Tu Anh Dinh、Felipe Maia Polo、Jan Niehues、Mrinmaya SachanComments: Maike Züfle, Vilém Zouhar, and Tu Anh Dinh contributed equally 注：Maike Züfle、Vilém Zouhar 和 Tu Anh Dinh 贡献相同Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [190] arXiv:2508.18473 [pdf, html, other]

  Principled Detection of Hallucinations in Large Language Models via Multiple Testing 通过多重检验对大型语言模型幻觉进行有原则的检测Jiawei Li, Akshayaa Magesh, Venugopal V. Veeravalli Jiawei Li、Akshayaa Magesh、Venugopal V. VeeravalliComments: 16 pages 备注：16 页Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [191] arXiv:2508.18466 [pdf, html, other]

  Integrating gender inclusivity into large language models via instruction tuning 通过指令微调将性别包容性融入大型语言模型Alina Wróblewska, Bartosz Żuk Alina Wróblewska，Bartosz ŻukSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [192] arXiv:2508.18444 [pdf, html, other]

  How Reliable are LLMs for Reasoning on the Re-ranking task? 大型语言模型在重排序任务的推理能力有多可靠？Nafis Tanveer Islam, Zhiming ZhaoComments: Accepted at FQAS Conference 2024. DOI will be provided in 3 weeks after the conference has published the paper 注释：已被 2024 年 FQAS 会议接受。会议发表论文后 3 周内将提供 DOISubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [193] arXiv:2508.18407 [pdf, html, other]

  Can Out-of-Distribution Evaluations Uncover Reliance on Shortcuts? A Case Study in Question Answering 是否可以通过分布外评估发现对捷径的依赖？在问答中的案例研究Michal Štefánik, Timothee Mickus, Marek Kadlčík, Michal Spiegel, Josef KuchařComments: To appear in Findings of EMNLP 2025 评注：将发表于 EMNLP 2025 FindingsSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [194] arXiv:2508.18395 [pdf, html, other]

  Latent Self-Consistency for Reliable Majority-Set Selection in Short- and Long-Answer Reasoning 潜在自洽性用于短回答和长回答推理中可靠的多数集选择Jeong-seok Oh, Jay-yoon Lee Jeong-seok Oh，Jay-yoon LeeSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [195] arXiv:2508.18387 [pdf, html, other]

  Integral Transformer: Denoising Attention, Not Too Much Not Too Little 积分变换器：去噪注意力，不多不少Ivan Kobyzev, Abbas Ghaddar, Dingtao Hu, Boxing Chen 伊万·科比泽夫、阿巴斯·加达尔、丁涛·胡、陈博兴Comments: EMNLP 2025 Main 评论：EMNLP 2025 主会议Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [196] arXiv:2508.18384 [pdf, html, other]

  Backprompting: Leveraging Synthetic Production Data for Health Advice Guardrails 反向提示：利用合成产出数据为健康建议建立保护措施Kellen Tan Cheng, Anna Lisa Gentile, Chad DeLuca, Guang-Jie RenSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [197] arXiv:2508.18381 [pdf, other]

  Language-Specific Layer Matters: Efficient Multilingual Enhancement for Large Vision-Language Models 语言特定层很重要：面向大型视觉-语言模型的高效多语言增强Yuchun Fan, Yilin Wang, Yongyu Mu, Lei Huang, Bei Li, Xiaocheng Feng, Tong Xiao, Jingbo Zhu 樊宇春，王一琳，穆咏宇，黄磊，李蓓，冯孝成，肖彤，祝靖博Comments: Accepted by EMNLP 2025 findings 评注：被 EMNLP 2025 findings 接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [198] arXiv:2508.18328 [pdf, html, other]

  Not All Visitors are Bilingual: A Measurement Study of the Multilingual Web from an Accessibility Perspective 并非所有访问者都会双语：从可及性视角对多语言网络的测量研究Masudul Hasan Masud Bhuiyan, Matteo Varvello, Yasir Zaki, Cristian-Alexandru Staicu Masudul Hasan Masud Bhuiyan、Matteo Varvello、Yasir Zaki、Cristian-Alexandru StaicuComments: 6 pages, 6 figures 评论：6 页，6 图Subjects: Computation and Language (cs.CL); Computers and Society (cs.CY); Networking and Internet Architecture (cs.NI) 学科：计算与语言 (cs.CL)；计算机与社会 (cs.CY)；网络与互联网架构 (cs.NI)

• [199] arXiv:2508.18321 [pdf, html, other]

  LLMs Can’t Handle Peer Pressure: Crumbling under Multi-Agent Social Interactions LLMs 无法应对同伴压力：在多智能体社交互动中崩溃Maojia Song, Tej Deep Pala, Weisheng Jin, Amir Zadeh, Chuan Li, Dorien Herremans, Soujanya PoriaSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [200] arXiv:2508.18290 [pdf, html, other]

  Semantic Attractors and the Emergence of Meaning: Towards a Teleological Model of AGI 语义吸引子与意义的涌现：走向一种目的论的通用人工智能模型Hans-Joachim Rudolph 汉斯-约阿希姆·鲁道夫Comments: 10 pages 备注：10 页Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [201] arXiv:2508.19229 (cross-list from cs.AI) [pdf, other] [201] arXiv:2508.19229（从 cs.AI 交叉列出）[ pdf，其他]

  StepWiser: Stepwise Generative Judges for Wiser Reasoning StepWiser：用于更明智推理的分步生成式评判器Wei Xiong, Wenting Zhao, Weizhe Yuan, Olga Golovneva, Tong Zhang, Jason Weston, Sainbayar Sukhbaatar 熊威，赵文婷，袁维哲，Olga Golovneva，张彤，Jason Weston，Sainbayar SukhbaatarSubjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [202] arXiv:2508.19200 (cross-list from cs.AI) [pdf, html, other] [202] arXiv:2508.19200（从 cs.AI 交叉列出）[ pdf, html, other]

  The Ramon Llull’s Thinking Machine for Automated Ideation Ramon Llull 的思维机器用于自动化创意生成Xinran Zhao, Boyuan Zheng, Chenglei Si, Haofei Yu, Ken Liu, Runlong Zhou, Ruochen Li, Tong Chen, Xiang Li, Yiming Zhang, Tongshuang Wu 赵欣然、郑博远、司承磊、于浩飞、Ken Liu、周润龙、李若晨、陈彤、李祥、张一鸣、吴通霜Comments: 21 pages, 3 figures 注：21 页，3 幅图Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [203] arXiv:2508.19005 (cross-list from cs.AI) [pdf, other] [203] arXiv:2508.19005（从 cs.AI 交叉列出）[ pdf，其他]

  Building Self-Evolving Agents via Experience-Driven Lifelong Learning: A Framework and Benchmark 通过经验驱动的终身学习构建自我进化智能体：一个框架与基准Yuxuan Cai, Yipeng Hao, Jie Zhou, Hang Yan, Zhikai Lei, Rui Zhen, Zhenhua Han, Yutao Yang, Junsong Li, Qianjun Pan, Tianyu Huai, Qin Chen, Xin Li, Kai Chen, Bo Zhang, Xipeng Qiu, Liang He 蔡煜轩，郝一鹏，周杰，闫航，雷志凯，郑睿，韩振华，杨玉涛，李俊松，潘千军，怀天宇，陈沁，李昕，陈凯，张博，邱锡鹏，何良Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [204] arXiv:2508.18976 (cross-list from cs.CR) [pdf, html, other] [204] arXiv:2508.18976（从 cs.CR 跨列）[ pdf，html，other ]

  The Double-edged Sword of LLM-based Data Reconstruction: Understanding and Mitigating Contextual Vulnerability in Word-level Differential Privacy Text Sanitization 基于 LLM 的数据重构的双刃剑：理解并缓解词级差分隐私文本清洗中的上下文脆弱性Stephen Meisenbacher, Alexandra Klymenko, Andreea-Elena Bodea, Florian Matthes Stephen Meisenbacher、Alexandra Klymenko、Andreea-Elena Bodea、Florian MatthesComments: 15 pages, 4 figures, 8 tables. Accepted to WPES @ CCS 2025 备注：15 页，4 幅图，8 张表。已被 WPES @ CCS 2025 接收Subjects: Cryptography and Security (cs.CR); Computation and Language (cs.CL) 主题：密码学与安全（cs.CR）；计算与语言（cs.CL）

• [205] arXiv:2508.18772 (cross-list from cs.CV) [pdf, other] [205] arXiv:2508.18772（从 cs.CV 交叉列出）[ pdf，其他]

  Beyond the Textual: Generating Coherent Visual Options for MCQs 超越文本：为选择题生成连贯的视觉选项Wanqiang Wang, Longzhu He, Wei Zheng 王万强，何龙珠，郑伟Comments: EMNLP 2025 评论：EMNLP 2025Subjects: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL) 主题：计算机视觉与模式识别 (cs.CV)；计算与语言 (cs.CL)

• [206] arXiv:2508.18760 (cross-list from cs.AI) [pdf, html, other] [206] arXiv:2508.18760（从 cs.AI 交叉列出）[ pdf，html，其他]

  Answering the Unanswerable Is to Err Knowingly: Analyzing and Mitigating Abstention Failures in Large Reasoning Models 有意回答无法回答的问题就是故意犯错：分析并缓解大型推理模型的弃答失败Yi Liu, Xiangyu Liu, Zequn Sun, Wei Hu 刘毅, 刘翔宇, 孙泽群, 胡伟Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [207] arXiv:2508.18758 (cross-list from cs.DB) [pdf, html, other] [207] arXiv:2508.18758（从 cs.DB 交叉列出）[ pdf, html, other]

  Text to Query Plans for Question Answering on Large Tables 用于大表格问答的查询计划文本Yipeng Zhang, Chen Wang, Yuzhe Zhang, Jacky Jiang 张一鹏，王晨，张宇哲，Jacky JiangSubjects: Databases (cs.DB); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：数据库（cs.DB）；人工智能（cs.AI）；计算与语言（cs.CL）

• [208] arXiv:2508.18743 (cross-list from cs.AI) [pdf, html, other] [208] arXiv:2508.18743（从 cs.AI 交叉收录）[ pdf, html, other]

  CAC-CoT: Connector-Aware Compact Chain-of-Thought for Efficient Reasoning Data Synthesis Across Dual-System Cognitive Tasks CAC-CoT：面向连接器的紧凑型链式思维，用于在双系统认知任务中高效合成推理数据Sunguk Choi, Yonghoon Kwon, Heondeuk Lee 崔成旭，权永勋，李憲德Comments: Accepted at EMNLP 2025 findings 备注：已被 EMNLP 2025 findings 接收Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [209] arXiv:2508.18724 (cross-list from cs.AI) [pdf, html, other] [209] arXiv:2508.18724（来自 cs.AI 的交叉投稿）[ pdf, html, other]

  Bias Mitigation Agent: Optimizing Source Selection for Fair and Balanced Knowledge Retrieval 偏见缓解代理：优化来源选择以实现公平和平衡的知识检索Karanbir Singh, Deepak Muppiri, William NguComments: Accepted at KDD'2025 Agent4IR workshop 评论：已被 KDD 2025 Agent4IR 研讨会接收Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [210] arXiv:2508.18684 (cross-list from cs.CR) [pdf, html, other] [210] arXiv:2508.18684（从 cs.CR 交叉列出）[ pdf, html, other]

  FALCON: Autonomous Cyber Threat Intelligence Mining with LLMs for IDS Rule Generation FALCON：使用 LLMs 进行入侵检测规则生成的自主网络威胁情报挖掘Shaswata Mitra, Azim Bazarov, Martin Duclos, Sudip Mittal, Aritran Piplai, Md Rayhanur Rahman, Edward Zieglar, Shahram Rahimi Shaswata Mitra、Azim Bazarov、Martin Duclos、Sudip Mittal、Aritran Piplai、Md Rayhanur Rahman、Edward Zieglar、Shahram RahimiComments: 11 pages, 5 figures, 4 tables 评论：11 页，5 张图，4 张表Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG); Systems and Control (eess.SY) 学科：密码学与安全（cs.CR）；人工智能（cs.AI）；计算与语言（cs.CL）；机器学习（cs.LG）；系统与控制（eess.SY）

• [211] arXiv:2508.18672 (cross-list from cs.LG) [pdf, html, other] [211] arXiv:2508.18672（来自 cs.LG 的交叉列表）[ pdf, html, other]

  Optimal Sparsity of Mixture-of-Experts Language Models for Reasoning Tasks 混合专家语言模型在推理任务中的最优稀疏性Taishi Nakamura, Satoki Ishikawa, Masaki Kawamura, Takumi Okamoto, Daisuke Nohara, Jun Suzuki, Rio Yokota 中村泰志、石川里央、河村正樹、冈本拓海、野原大辅、铃木淳、横田理央Comments: Presented at the Second AI for Math Workshop at ICML 备注：在 ICML 举办的第二届 AI for Math 研讨会上展示Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：机器学习 (cs.LG); 人工智能 (cs.AI); 计算与语言 (cs.CL)

• [212] arXiv:2508.18665 (cross-list from cs.IR) [pdf, html, other] [212] arXiv:2508.18665（从 cs.IR 交叉列出）[ pdf, html, other]

  Membership Inference Attacks on LLM-based Recommender Systems 基于 LLM 的推荐系统上的成员推断攻击Jiajie He, Yuechun Gu, Min-Chun Chen, Keke Chen 贺佳杰，顾越春，陈敏纯，陈可可Subjects: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Cryptography and Security (cs.CR); Machine Learning (cs.LG) 主题：信息检索（cs.IR）；人工智能（cs.AI）；计算与语言（cs.CL）；密码学与安全（cs.CR）；机器学习（cs.LG）

• [213] arXiv:2508.18652 (cross-list from cs.CR) [pdf, html, other] [213] arXiv:2508.18652（从 cs.CR 交叉列出）[ pdf, html, other]

  UniC-RAG: Universal Knowledge Corruption Attacks to Retrieval-Augmented Generation UniC-RAG：面向检索增强生成的通用知识破坏攻击Runpeng Geng, Yanting Wang, Ying Chen, Jinyuan Jia 耿润鹏，王艳婷，陈颖，贾金元Comments: 21 pages, 4 figures 注释：21 页，4 幅图Subjects: Cryptography and Security (cs.CR); Computation and Language (cs.CL) 主题：密码学与安全（cs.CR）；计算与语言（cs.CL）

• [214] arXiv:2508.18646 (cross-list from cs.AI) [pdf, html, other] [214] arXiv:2508.18646（跨列自 cs.AI）[ pdf，html，other]

  Beyond Benchmark: LLMs Evaluation with an Anthropomorphic and Value-oriented Roadmap 超越基准：基于拟人化与价值导向路线图的 LLMs 评估Jun Wang, Ninglun Gu, Kailai Zhang, Zijiao Zhang, Yelun Bao, Jin Yang, Xu Yin, Liwei Liu, Yihuan Liu, Pengyong Li, Gary G. Yen, Junchi Yan 王军，顾宁伦，张凯来，张子娇，鲍叶伦，杨晋，尹旭，刘立伟，刘亦焕，李朋勇，Gary G. Yen，严俊驰Comments: Preprint. Under review 评论：预印本。审稿中Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [215] arXiv:2508.18642 (cross-list from cs.AI) [pdf, html, other] [215] arXiv:2508.18642（跨榜单自 cs.AI）[ pdf，html，其他]

  RLMR: Reinforcement Learning with Mixed Rewards for Creative Writing RLMR：用于创意写作的混合奖励强化学习Jianxing Liao, Tian Zhang, Xiao Feng, Yusong Zhang, Rui Yang, Haorui Wang, Bosi Wen, Ziying Wang, Runzhi Shi 廖建兴、张天、冯啸、张宇松、杨锐、王昊睿、温博思、王梓英、史润志Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [216] arXiv:2508.18512 (cross-list from physics.optics) [pdf, html, other] [216] arXiv:2508.18512（从 physics.optics 交叉列出）[ pdf, html, other]

  Designing across domains with declarative thinking: Insights from the 96-Eyes ptychographic imager project 以声明式思维进行跨域设计：来自“96 眼相位恢复成像仪”项目的见解Antony C Chan 安东尼·C·陈Subjects: Optics (physics.optics); Computation and Language (cs.CL) 主题：光学（physics.optics）；计算与语言（cs.CL）

• [217] arXiv:2508.18439 (cross-list from cs.CR) [pdf, html, other] [217] arXiv:2508.18439（由 cs.CR 交叉列出）[ pdf，html，other]

  A Systematic Approach to Predict the Impact of Cybersecurity Vulnerabilities Using LLMs 使用 LLMs 系统性预测网络安全漏洞影响的方法Anders Mølmen Høst, Pierre Lison, Leon Moonen Anders Mølmen Høst，Pierre Lison，Leon MoonenSubjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Software Engineering (cs.SE) 主题：密码学与安全（cs.CR）；人工智能（cs.AI）；计算与语言（cs.CL）；软件工程（cs.SE）

• [218] arXiv:2508.18370 (cross-list from cs.SE) [pdf, other] [218] arXiv:2508.18370（从 cs.SE 交叉列出）[ pdf，其他]

  Training Language Model Agents to Find Vulnerabilities with CTF-Dojo 训练语言模型代理在 CTF-Dojo 中发现漏洞Terry Yue Zhuo, Dingmin Wang, Hantian Ding, Varun Kumar, Zijian Wang Terry Yue Zhuo、Dingmin Wang、Hantian Ding、Varun Kumar、Zijian WangSubjects: Software Engineering (cs.SE); Computation and Language (cs.CL); Cryptography and Security (cs.CR); Machine Learning (cs.LG) 主题：软件工程 (cs.SE); 计算与语言 (cs.CL); 密码学与安全 (cs.CR); 机器学习 (cs.LG)

• [219] arXiv:2508.18306 (cross-list from cs.LG) [pdf, html, other] [219] arXiv:2508.18306（来自 cs.LG 的交叉列出）[ pdf, html, other]

  SALMAN: Stability Analysis of Language Models Through the Maps Between Graph-based Manifolds SALMAN：通过基于图的流形之间的映射对语言模型的稳定性分析Wuxinlin Cheng, Yupeng Cao, Jinwen Wu, Koduvayur Subbalakshmi, Tian Han, Zhuo Feng 程无忻，曹宇鹏，吴金文，Koduvayur Subbalakshmi，韩天，冯卓Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：机器学习 (cs.LG); 人工智能 (cs.AI); 计算与语言 (cs.CL)

• [220] arXiv:2508.18297 (cross-list from cs.CV) [pdf, html, other] [220] arXiv:2508.18297（跨列表自 cs.CV）[ pdf, html, 其他]

  Can VLMs Recall Factual Associations From Visual References? 视觉语言模型能否从视觉参考中回忆起事实性关联？Dhananjay Ashok, Ashutosh Chaubey, Hirona J. Arai, Jonathan May, Jesse Thomason Dhananjay Ashok、Ashutosh Chaubey、Hirona J. Arai、Jonathan May、Jesse ThomasonComments: To appear at EMNLP 2025 (Findings) 评论：将发表于 EMNLP 2025（Findings）Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：计算机视觉与模式识别（cs.CV）；人工智能（cs.AI）；计算与语言（cs.CL）

• [221] arXiv:2508.18295 (cross-list from cs.SD) [pdf, html, other] [221] arXiv:2508.18295（从 cs.SD 交叉列出）[ pdf、html、other]

  H-PRM: A Pluggable Hotword Pre-Retrieval Module for Various Speech Recognition Systems H-PRM：一种可插拔的热词检索前模块，适用于各种语音识别系统Huangyu Dai, Lingtao Mao, Ben Chen, Zihan Wang, Zihan Liang, Ying Han, Chenyi Lei, Han Li 黄宇岱、毛灵韬、陈本、王梓涵、梁梓涵、韩颖、雷晨毅、李涵Subjects: Sound (cs.SD); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Audio and Speech Processing (eess.AS) 主题：声音 (cs.SD)；人工智能 (cs.AI)；计算与语言 (cs.CL)；音频与语音处理 (eess.AS)

• [222] arXiv:2508.18288 (cross-list from eess.AS) [pdf, other] [222] arXiv:2508.18288（从 eess.AS 交叉列出）[ pdf, other]

  Toward Responsible ASR for African American English Speakers: A Scoping Review of Bias and Equity in Speech Technology 面向非裔美国英语使用者的负责任自动语音识别：语音技术中偏见与公平性的范围综述Jay L. Cunningham, Adinawa Adjagbodjou, Jeffrey Basoah, Jainaba Jawara, Kowe Kadoma, Aaleyah LewisComments: 10 pages, 9 Pages (References and Appendices). The archival version has been accepted to AAAI (AIES 2025) without the extended Appendices. This extended version includes Appendices 备注：10 页，参考文献和附录共 9 页。存档版本已被 AAAI（AIES 2025）接收，未包含扩展附录。本扩展版本包含附录Subjects: Audio and Speech Processing (eess.AS); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Sound (cs.SD) 主题：音频与语音处理 (eess.AS)；人工智能 (cs.AI)；计算与语言 (cs.CL)；声音 (cs.SD)

• [223] arXiv:2508.18260 [pdf, html, other]

  MIRAGE: Scaling Test-Time Inference with Parallel Graph-Retrieval-Augmented Reasoning Chains MIRAGE：通过并行图检索增强推理链扩展测试时推理规模Kaiwen Wei, Rui Shan, Dongsheng Zou, Jianzhong Yang, Bi Zhao, Junnan Zhu, Jiang Zhong 魏凯文，单睿，邹东升，杨建中，赵碧，朱俊南，钟江Comments: 10 pages, 8 figures (including tables), plus appendix. Submitted to AAAI 2026 注释：10 页，8 幅图（含表格），另附附录。已提交至 AAAI 2026Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [224] arXiv:2508.18253 [pdf, html, other]

  From BERT to LLMs: Comparing and Understanding Chinese Classifier Prediction in Language Models 从 BERT 到 LLMs：比较与理解语言模型中中文分类器的预测ZiqiZhang, Jianfei Ma, Emmanuele Chersoni, Jieshun You, Zhaoxin Feng 张子齐，马建飞，埃曼努埃莱·凯尔索尼，游杰顺，冯昭昕Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [225] arXiv:2508.18245 [pdf, html, other]

  Demographic Biases and Gaps in the Perception of Sexism in Large Language Models 大型语言模型在性别歧视感知方面的人口统计偏见与差距Judith Tavarez-Rodríguez, Fernando Sánchez-Vega, A. Pastor López-MonroyComments: This work was presented as a poster at the Latin American Meeting in Artificial Intelligence KHIPU 2025, Santiago, Chile, March 10th - 14th 2025, this https URL 备注：本工作以海报形式在 2025 年 3 月 10 日至 14 日于智利圣地亚哥召开的拉丁美洲人工智能会议 KHIPU 2025 上展示，网址为 this https URLSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [226] arXiv:2508.18240 [pdf, html, other]

  MTalk-Bench: Evaluating Speech-to-Speech Models in Multi-Turn Dialogues via Arena-style and Rubrics Protocols MTalk-Bench：通过竞技场式与量表协议在多轮对话中评估语音到语音模型Yuhao Du, Qianwei Huang, Guo Zhu, Zhanchen Dai, Sunian Chen, Qiming Zhu, Yuhao Zhang, Li Zhou, Benyou Wang 杜宇豪、黄千渭、祝国、戴展宸、陈素年、朱启明、张宇豪、周莉、王本有Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [227] arXiv:2508.18212 [pdf, html, other]

  Better Language Model-Based Judging Reward Modeling through Scaling Comprehension Boundaries 通过扩展理解边界改进基于更好语言模型的评分奖励建模Meiling Ning, Zhongbao Zhang, Junda Ye, Jiabao Guo, Qingyuan Guan 宁美玲，张中宝，叶俊达，郭嘉宝，管庆元Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [228] arXiv:2508.18210 [pdf, html, other]

  Why Synthetic Isn’t Real Yet: A Diagnostic Framework for Contact Center Dialogue Generation 为什么合成还不是真实的：呼叫中心对话生成的诊断框架Rishikesh Devanathan, Varun Nathan, Ayush Kumar 里希凯什·德瓦纳坦，瓦伦·纳森，阿育什·库马尔Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [229] arXiv:2508.18208 [pdf, html, other]

  Exploring the Interplay between Musical Preferences and Personality through the Lens of Language 通过语言视角探索音乐偏好与人格之间的相互作用Eliran Shem-Tov, Ella Rabinovich 埃利兰·谢姆托夫，艾拉·拉比诺维奇Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [230] arXiv:2508.18183 [pdf, html, other]

  Leveraging Large Language Models for Accurate Sign Language Translation in Low-Resource Scenarios 在低资源场景中利用大型语言模型实现准确的手语翻译Luana Bulla, Gabriele Tuccio, Misael Mongiovì, Aldo Gangemi Luana Bulla、Gabriele Tuccio、Misael Mongiovì、Aldo GangemiSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；计算机与社会 (cs.CY)

• [231] arXiv:2508.18168 [pdf, html, other]

  Improving End-to-End Training of Retrieval-Augmented Generation Models via Joint Stochastic Approximation 通过联合随机逼近改进增强检索生成模型的端到端训练Hongyu Cao, Yuxuan Wu, Yucheng Cai, Xianyu Zhao, Zhijian Ou 曹鸿宇，吴宇轩，蔡雨程，赵仙玉，欧志坚Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [232] arXiv:2508.18167 [pdf, other]

  DiscussLLM: Teaching Large Language Models When to Speak DiscussLLM：教大型语言模型何时该发言Deep Anil Patel, Iain Melvin, Christopher Malon, Martin Renqiang Min Deep Anil Patel，Iain Melvin，Christopher Malon，Martin Renqiang MinSubjects: Computation and Language (cs.CL); Human-Computer Interaction (cs.HC) 学科：计算与语言（cs.CL）；人机交互（cs.HC）

• [233] arXiv:2508.18164 [pdf, html, other]

  S2Sent: Nested Selectivity Aware Sentence Representation Learning S2Sent：嵌套选择性感知的句子表示学习Jianxiang Zang, Nijia Mo, Yonda Wei, Meiling Ning, Hui Liu 臧建祥，莫妮佳，魏源达，宁美玲，刘辉Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [234] arXiv:2508.18134 [pdf, other] [234] arXiv:2508.18134 [ pdf，其他]

  Toward a Better Localization of Princeton WordNet 走向对普林斯顿词网更好的本地化Abed Alhakim Freihat 阿贝德·阿尔哈基姆·弗赖哈特Comments: in Arabic language 评论：用阿拉伯语Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [235] arXiv:2508.18108 [pdf, html, other]

  SentiMM: A Multimodal Multi-Agent Framework for Sentiment Analysis in Social Media SentiMM：一个用于社交媒体情感分析的多模态多代理框架Xilai Xu, Zilin Zhao, Chengye Song, Zining Wang, Jinhe Qiang, Jiongrui Yan, Yuhuai LinSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [236] arXiv:2508.18098 [pdf, html, other]

  Detecting and Characterizing Planning in Language Models 在语言模型中检测与刻画规划Jatin Nainani, Sankaran Vaidyanathan, Connor Watts, Andre N. Assis, Alice Rigg Jatin Nainani、Sankaran Vaidyanathan、Connor Watts、Andre N. Assis、Alice RiggComments: 9 pages, 4 figures 注释：9 页，4 图Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [237] arXiv:2508.18093 [pdf, other]

  Agri-Query: A Case Study on RAG vs. Long-Context LLMs for Cross-Lingual Technical Question Answering Agri-Query：关于用于跨语种技术问答的 RAG 与长上下文 LLMs 的案例研究Julius Gun, Timo Oksanen Julius Gun，Timo OksanenSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [238] arXiv:2508.18092 [pdf, html, other]

  Speech-Based Depressive Mood Detection in the Presence of Multiple Sclerosis: A Cross-Corpus and Cross-Lingual Study 基于语音的抑郁情绪检测在多发性硬化存在下的研究：一项跨语料库与跨语种研究Monica Gonzalez-Machorro, Uwe Reichel, Pascal Hecker, Helly Hammer, Hesam Sagha, Florian Eyben, Robert Hoepner, Björn W. Schuller 莫妮卡·冈萨雷斯-马乔罗, 乌韦·赖歇尔, 帕斯卡尔·赫克, 赫莉·哈默, 赫萨姆·萨加, 弗洛里安·艾本, 罗伯特·霍普纳, 比约恩·W. 舒勒Comments: Accepted at the 8th International Conference on Natural Language and Speech Processing (ICNLSP 2025). To be appeared in the corresponding Proceedings at ACL Anthology 评注：已被第八届国际自然语言与语音处理会议（ICNLSP 2025）接收。将收录于 ACL Anthology 对应的会议论文集Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [239] arXiv:2508.18088 [pdf, other]

  How Quantization Shapes Bias in Large Language Models 量化如何影响大型语言模型中的偏差Federico Marcuzzi, Xuefei Ning, Roy Schwartz, Iryna GurevychSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [240] arXiv:2508.18076 [pdf, html, other]

  Neither Valid nor Reliable? Investigating the Use of LLMs as Judges 既不可信也不可靠？调查将 LLMs 用作裁判的做法Khaoula Chehbouni, Mohammed Haddou, Jackie Chi Kit Cheung, Golnoosh Farnadi Khaoula Chehbouni、Mohammed Haddou、Jackie Chi Kit Cheung、Golnoosh FarnadiComments: Prepared for conference submission 注释：为会议提交准备Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [241] arXiv:2508.17994 [pdf, html, other]

  A Retail-Corpus for Aspect-Based Sentiment Analysis with Large Language Models 一个用于面向方面情感分析并适配大型语言模型的零售语料库Oleg Silcenco, Marcos R. Machad, Wallace C. Ugulino, Daniel BraunComments: Accepted at ICNLSP 2025 备注：已被 ICNLSP 2025 接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [242] arXiv:2508.17973 [pdf, html, other]

  German4All - A Dataset and Model for Readability-Controlled Paraphrasing in German German4All - 一个用于德语可读性控制意译的数据集和模型Miriam Anschütz, Thanh Mai Pham, Eslam Nasrallah, Maximilian Müller, Cristian-George Craciun, Georg Groh Miriam Anschütz、Thanh Mai Pham、Eslam Nasrallah、Maximilian Müller、Cristian-George Craciun、Georg GrohComments: Accepted to INLG 2025 注释：已被 INLG 2025 接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [243] arXiv:2508.17953 [pdf, html, other]

  Understanding Subword Compositionality of Large Language Models 理解大型语言模型的子词组合性Qiwei Peng, Yekun Chai, Anders Søgaard 彭祺威、柴业坤、安德斯·索格德Comments: EMNLP 2025 Main 评论：EMNLP 2025 主会议Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [244] arXiv:2508.17948 [pdf, html, other]

  Debiasing Multilingual LLMs in Cross-lingual Latent Space 在跨语言潜在空间中对多语种 LLMs 进行去偏Qiwei Peng, Guimin Hu, Yekun Chai, Anders Søgaard 彭祺威、胡桂敏、柴业坤、安德斯·索格德Comments: EMNLP 2025 Main 评论：EMNLP 2025 主会议Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [245] arXiv:2508.17926 [pdf, html, other]

  AMELIA: A Family of Multi-task End-to-end Language Models for Argumentation AMELIA：一系列用于论证的多任务端到端语言模型Henri Savigny, Bruno Yun 亨利·萨维尼，布鲁诺·云Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [246] arXiv:2508.17923 [pdf, other]

  Feature-Refined Unsupervised Model for Loanword Detection 基于特征精炼的无监督借词检测模型Promise Dodzi KpogluSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [247] arXiv:2508.17918 [pdf, other]

  Information availability in different languages and various technological constraints related to multilinguism on the Internet 不同语言的信息可用性以及与互联网多语言性相关的各种技术限制Sonal Khosla, Haridasa Acharya Sonal Khosla、Haridasa AcharyaComments: International Journal of Computer Applications 评论：国际计算机应用期刊Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [248] arXiv:2508.17914 [pdf, html, other]

  Evaluating the Representation of Vowels in Wav2Vec Feature Extractor: A Layer-Wise Analysis Using MFCCs 评估 Wav2Vec 特征提取器中元音的表征：使用 MFCC 的逐层分析Domenico De Cristofaro, Vincenzo Norman Vitale, Alessandro ViettiSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [249] arXiv:2508.17905 [pdf, html, other]

  Pandora: Leveraging Code-driven Knowledge Transfer for Unified Structured Knowledge Reasoning Pandora：利用代码驱动的知识迁移实现统一的结构化知识推理Yongrui Chen, Junhao He, Linbo Fu, Shenyu Zhang, Rihui Jin, Xinbang Dai, Jiaqi Li, Dehai Min, Nan Hu, Yuxin Zhang, Guilin Qi, Yi Huang, Tongtong Wu 陈永睿、何俊豪、傅林博、张神宇、金日辉、戴新邦、李佳琦、闵德海、胡楠、张宇鑫、齐桂林、黄怡、吴彤彤Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [250] arXiv:2508.17892 [pdf, html, other]

  ILRe: Intermediate Layer Retrieval for Context Compression in Causal Language Models ILRe：因果语言模型中用于上下文压缩的中间层检索Manlai Liang, Mandi Liu, Jiangzhou Ji, Huaijun Li, Haobo Yang, Yaohan He, Jinlong Li 梁曼来、刘蔓迪、纪江洲、李怀军、杨浩博、何遥涵、李金龙Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [251] arXiv:2508.17863 [pdf, html, other]

  Speech Discrete Tokens or Continuous Features? A Comparative Analysis for Spoken Language Understanding in SpeechLLMs 语音：离散标记还是连续特征？在 SpeechLLMs 中用于语音语言理解的比较分析Dingdong Wang, Junan Li, Mingyu Cui, Dongchao Yang, Xueyuan Chen, Helen Meng 王叮咚，李俊安，崔明宇，杨东超，陈学元，孟海伦Comments: Accepted to EMNLP 2025 Main Conference 注释：已被接收为 EMNLP 2025 主会会议论文Subjects: Computation and Language (cs.CL); Sound (cs.SD) 主题：计算与语言（cs.CL）；声音（cs.SD）

• [252] arXiv:2508.17855 [pdf, html, other]

  Beyond Demographics: Enhancing Cultural Value Survey Simulation with Multi-Stage Personality-Driven Cognitive Reasoning 超越人口统计：通过多阶段以人格为驱动的认知推理提升文化价值调查模拟Haijiang Liu, Qiyuan Li, Chao Gao, Yong Cao, Xiangyu Xu, Xun Wu, Daniel Hershcovich, Jinguang Gu 刘海江、李其元、高超、曹勇、徐向宇、吴勋、Daniel Hershcovich、古劲光Comments: 23 pages, 6 figures, accepted to EMNLP 2025 main 注释：23 页，6 幅图，已被接受为 EMNLP 2025 主要会议论文Subjects: Computation and Language (cs.CL); Computers and Society (cs.CY) 学科：计算与语言（cs.CL）；计算机与社会（cs.CY）

• [253] arXiv:2508.17803 [pdf, html, other]

  DRQA: Dynamic Reasoning Quota Allocation for Controlling Overthinking in Reasoning Large Language Models DRQA：用于控制推理大型语言模型过度思考的动态推理配额分配Kaiwen Yan, Xuanqing Shi, Hongcheng Guo, Wenxuan Wang, Zhuosheng Zhang, Chengwei Qin 闫凯文、史轩清、郭宏成、王文轩、张卓胜、覃成伟Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [254] arXiv:2508.17796 [pdf, html, other]

  Zero-shot Context Biasing with Trie-based Decoding using Synthetic Multi-Pronunciation 使用合成多发音的基于字典树解码的零样本上下文偏置Changsong Liu, Yizhou Peng, Eng Siong ChngComments: Accepted to APSIPA ASC 2025 备注：已被 APSIPA ASC 2025 接收Subjects: Computation and Language (cs.CL); Audio and Speech Processing (eess.AS) 主题：计算与语言（cs.CL）；音频与语音处理（eess.AS）

• [255] arXiv:2508.17771 [pdf, html, other]

  Speculating LLMs’ Chinese Training Data Pollution from Their Tokens 从其标记推测 LLMs 的中文训练数据污染情况Qingjie Zhang, Di Wang, Haoting Qian, Liu Yan, Tianwei Zhang, Ke Xu, Qi Li, Minlie Huang, Hewu Li, Han Qiu 张庆杰、王迪、钱皓廷、柳嫣、张天伟、徐可、李琦、黄民烈、李和武、裘涵Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [256] arXiv:2508.17767 [pdf, html, other]

  ISACL: Internal State Analyzer for Copyrighted Training Data Leakage ISACL：用于有版权训练数据泄露的内部状态分析器Guangwei Zhang, Qisheng Su, Jiateng Liu, Cheng Qian, Yanzhou Pan, Yanjie Fu, Denghui Zhang 张广伟，苏启盛，刘佳腾，钱成，潘艳周，付岩杰，张登辉Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [257] arXiv:2508.17735 [pdf, html, other]

  SMITE: Enhancing Fairness in LLMs through Optimal In-Context Example Selection via Dynamic Validation SMITE：通过动态验证的最优上下文示例选择提升 LLMs 的公平性Garima Chhikara, Kripabandhu Ghosh, Abhijnan Chakraborty Garima Chhikara、Kripabandhu Ghosh、Abhijnan ChakrabortySubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [258] arXiv:2508.17734 [pdf, html, other]

  Layerwise Importance Analysis of Feed-Forward Networks in Transformer-based Language Models 基于 Transformer 的语言模型中前馈网络的层级重要性分析Wataru Ikeda, Kazuki Yano, Ryosuke Takahashi, Jaesung Lee, Keigo Shibata, Jun Suzuki Wataru Ikeda、Kazuki Yano、Ryosuke Takahashi、Jaesung Lee、Keigo Shibata、Jun SuzukiComments: Accepted to COLM 2025 注释：已被 COLM 2025 接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [259] arXiv:2508.17703 [pdf, html, other]

  EMPOWER: Evolutionary Medical Prompt Optimization With Reinforcement Learning EMPOWER：结合强化学习的进化医学提示优化Yinda Chen, Yangfan He, Jing Yang, Dapeng Zhang, Zhenlong Yuan, Muhammad Attique Khan, Jamel Baili, Por Lip YeeSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [260] arXiv:2508.17690 [pdf, html, other]

  Text Meets Topology: Rethinking Out-of-distribution Detection in Text-Rich Networks 文本遇见拓扑：在富文本网络中重新思考分布外检测Danny Wang, Ruihong Qiu, Guangdong Bai, Zi HuangComments: EMNLP2025 Main 备注：EMNLP2025 主会稿Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [261] arXiv:2508.17670 [pdf, html, other]

  CoCoA: Confidence and Context-Aware Adaptive Decoding for Resolving Knowledge Conflicts in Large Language Models CoCoA：用于解决大型语言模型知识冲突的置信度与上下文感知自适应解码Anant Khandelwal, Manish Gupta, Puneet Agrawal Anant Khandelwal、Manish Gupta、Puneet AgrawalComments: Accepted to EMNLP'25, Main. 21 pages, 17 tables, 3 Figures 评注：已被 EMNLP'25 主会议接收。21 页，17 张表格，3 幅图Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [262] arXiv:2508.17647 [pdf, html, other]

  SurveyGen: Quality-Aware Scientific Survey Generation with Large Language Models SurveyGen：面向质量感知的大型语言模型科学综述生成Tong Bao, Mir Tafseer Nayeem, Davood Rafiei, Chengzhi Zhang Tong Bao、Mir Tafseer Nayeem、Davood Rafiei、Chengzhi ZhangJournal-ref: EMNLP2025 期刊参考：EMNLP2025Subjects: Computation and Language (cs.CL); Digital Libraries (cs.DL); Information Retrieval (cs.IR) 主题：计算与语言（cs.CL）；数字图书馆（cs.DL）；信息检索（cs.IR）

• [263] arXiv:2508.17637 [pdf, html, other]

  Weights-Rotated Preference Optimization for Large Language Models 用于大型语言模型的权重旋转偏好优化Chenxu Yang, Ruipeng Jia, Mingyu Zheng, Naibin Gu, Zheng Lin, Siyuan Chen, Weichong Yin, Hua Wu, Weiping Wang 陈旭阳、贾瑞鹏、郑明宇、顾乃斌、林正、陈思源、尹伟冲、华吴、王卫平Comments: EMNLP 2025 评论：EMNLP 2025Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [264] arXiv:2508.17627 [pdf, html, other]

  Stop Spinning Wheels: Mitigating LLM Overthinking via Mining Patterns for Early Reasoning Exit 停止原地打转：通过挖掘模式实现提前推理退出以缓解 LLM 过度思考Zihao Wei, Liang Pang, Jiahao Liu, Jingcheng Deng, Shicheng Xu, Zenghao Duan, Jingang Wang, Fei Sun, Xunliang Cai, Huawei Shen, Xueqi Cheng 魏子豪、庞亮、刘佳豪、邓景程、徐世成、段增浩、王金刚、孙飞、蔡勋良、沈华为、程学旗Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [265] arXiv:2508.17623 [pdf, html, other]

  EMO-Reasoning: Benchmarking Emotional Reasoning Capabilities in Spoken Dialogue Systems EMO-Reasoning：评估口语对话系统中的情感推理能力Jingwen Liu, Kan Jen Cheng, Jiachen Lian, Akshay Anand, Rishi Jain, Faith Qiao, Robin Netzorg, Huang-Cheng Chou, Tingle Li, Guan-Ting Lin, Gopala Anumanchipalli Jingwen Liu、Kan Jen Cheng、Jiachen Lian、Akshay Anand、Rishi Jain、Faith Qiao、Robin Netzorg、Huang-Cheng Chou、Tingle Li、Guan-Ting Lin、Gopala AnumanchipalliComments: Accepted at (ASRU 2025) 2025 IEEE Automatic Speech Recognition and Understanding Workshop 评注：已被 (ASRU 2025) 2025 年 IEEE 自动语音识别与理解研讨会接收Subjects: Computation and Language (cs.CL); Audio and Speech Processing (eess.AS) 主题：计算与语言（cs.CL）；音频与语音处理（eess.AS）

• [266] arXiv:2508.17621 [pdf, html, other]

  Steering When Necessary: Flexible Steering Large Language Models with Backtracking 必要时引导：通过回溯灵活引导大型语言模型Jinwei Gan, Zifeng Cheng, Zhiwei Jiang, Cong Wang, Yafeng Yin, Xiang Luo, Yuchen Fu, Qing Gu Jinwei Gan、Zifeng Cheng、Zhiwei Jiang、Cong Wang、Yafeng Yin、Xiang Luo、Yuchen Fu、Qing GuSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [267] arXiv:2508.17610 [pdf, html, other] [267] arXiv:2508.17610 [ pdf、html、other]

  Less Is More? Examining Fairness in Pruned Large Language Models for Summarising Opinions 更少就是更多？审视用于总结观点的剪枝大型语言模型中的公平性Nannan Huang, Haytham Fayek, Xiuzhen Zhang Nannan Huang、Haytham Fayek、Xiuzhen ZhangComments: Accepted to EMNLP 2025 Main Conference 注释：已被接收为 EMNLP 2025 主会会议论文Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [268] arXiv:2508.17580 [pdf, other]

  UQ: Assessing Language Models on Unsolved Questions UQ：在未解问题上评估语言模型Fan Nie, Ken Ziyu Liu, Zihao Wang, Rui Sun, Wei Liu, Weijia Shi, Huaxiu Yao, Linjun Zhang, Andrew Y. Ng, James Zou, Sanmi Koyejo, Yejin Choi, Percy Liang, Niklas Muennighoff 范聂、刘子煜、王子豪、孙睿、刘巍、史维佳、姚华秀、张林军、Andrew Y. Ng、James Zou、Sanmi Koyejo、Yejin Choi、Percy Liang、Niklas MuennighoffComments: FN, KZL, and NM are project co-leads and contributed equally. Project website: this https URL 备注：FN、KZL 和 NM 为项目共同负责人，贡献相同。项目网站：this https URLSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [269] arXiv:2508.17576 [pdf, html, other]

  CausalSent: Interpretable Sentiment Classification with RieszNet CausalSent：带有 RieszNet 的可解释情感分类Daniel Frees, Martin Pollack Daniel Frees、Martin PollackSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [270] arXiv:2508.17573 [pdf, html, other]

  Humanizing Machines: Rethinking LLM Anthropomorphism Through a Multi-Level Framework of Design 使机器更有人性：通过多层次设计框架重新审视对 LLM 的人格化Yunze Xiao, Lynnette Hui Xian Ng, Jiarui Liu, Mona T. Diab 肖云泽，林妮特·许仙·黄，刘嘉睿，莫娜·T·迪亚布Comments: Accepted in EMNLP main proceedings 备注：已被 EMNLP 主会论文接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [271] arXiv:2508.17536 [pdf, html, other]

  Debate or Vote: Which Yields Better Decisions in Multi-Agent Large Language Models? 辩论还是投票：在多智能体大语言模型中哪种方式能产生更好的决策？Hyeong Kyu Choi, Xiaojin Zhu, Yixuan Li 崔炯圭 (Hyeong Kyu Choi)、朱小谨 (Xiaojin Zhu)、李意轩 (Yixuan Li)Subjects: Computation and Language (cs.CL); Multiagent Systems (cs.MA) 主题：计算与语言（cs.CL）；多智能体系统（cs.MA）

• [272] arXiv:2508.17494 [pdf, html, other]

  Improving French Synthetic Speech Quality via SSML Prosody Control 通过 SSML 重音控制提升法语合成语音质量Nassima Ould Ouali, Awais Hussain Sani, Ruben Bueno, Jonah Dauvet, Tim Luka Horstmann, Eric MoulinesComments: 13 pages, 9 figures, 6 tables. Accepted for presentation at ICNLSP 2025 (Odense, Denmark). Code and demo: this https URL. ACM Class: I.2.7; H.5.5 注释：13 页，9 幅图，6 张表。已被接受在 ICNLSP 2025（丹麦欧登塞）上展示。代码与演示：此 https URL。ACM 分类：I.2.7；H.5.5Subjects: Computation and Language (cs.CL); Sound (cs.SD) 主题：计算与语言（cs.CL）；声音（cs.SD）

• [273] arXiv:2508.17490 [pdf, html, other]

  Efficient Zero-Shot Long Document Classification by Reducing Context Through Sentence Ranking 通过句子排序减少上下文实现高效零样本文档长文本分类Prathamesh Kokate, Mitali Sarnaik, Manavi Khopade, Mukta Takalikar, Raviraj JoshiSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [274] arXiv:2508.17458 [pdf, html, other]

  Evaluating the Impact of Verbal Multiword Expressions on Machine Translation 评估词语多词表达对机器翻译的影响Linfeng Liu, Saptarshi Ghosh, Tianyu Jiang Linfeng Liu、Saptarshi Ghosh、Tianyu JiangComments: 29 pages, 13 figures 备注：29 页，13 幅图Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [275] arXiv:2508.17450 [pdf, html, other]

  Persuasion Dynamics in LLMs: Investigating Robustness and Adaptability in Knowledge and Safety with DuET-PD 说服动态在 LLMs 中的研究：使用 DuET-PD 调查知识和安全性的稳健性与适应性Bryan Chen Zhengyu Tan, Daniel Wai Kit Chin, Zhengyuan Liu, Nancy F. Chen, Roy Ka-Wei LeeComments: To appear at EMNLP 2025 注释：将发表于 EMNLP 2025Subjects: Computation and Language (cs.CL); Computers and Society (cs.CY) 学科：计算与语言（cs.CL）；计算机与社会（cs.CY）

• [276] arXiv:2508.17444 [pdf, html, other]

  MahaParaphrase: A Marathi Paraphrase Detection Corpus and BERT-based Models MahaParaphrase：一个马拉地语复述检测语料库及基于 BERT 的模型Suramya Jadhav, Abhay Shanbhag, Amogh Thakurdesai, Ridhima Sinare, Ananya Joshi, Raviraj Joshi Suramya Jadhav、Abhay Shanbhag、Amogh Thakurdesai、Ridhima Sinare、Ananya Joshi、Raviraj JoshiSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [277] arXiv:2508.17402 [pdf, html, other]

  DS@GT at CheckThat! 2025: A Simple Retrieval-First, LLM-Backed Framework for Claim Normalization DS@GT 在 CheckThat! 2025：一种用于陈述规范化的简单检索优先、LLM 支持框架Aleksandar Pramov, Jiangqin Ma, Bina Patel Aleksandar Pramov，马江琴，Bina PatelComments: CLEF 2025 Working Notes, Madrid, Spain 评注：CLEF 2025 工作手记，西班牙，马德里Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；信息检索 (cs.IR)

• [278] arXiv:2508.17398 [pdf, html, other]

  DashboardQA: Benchmarking Multimodal Agents for Question Answering on Interactive Dashboards DashboardQA：用于交互式仪表盘问答的多模态智能体基准测试Aaryaman Kartha, Ahmed Masry, Mohammed Saidul Islam, Thinh Lang, Shadikur Rahman, Ridwan Mahbub, Mizanur Rahman, Mahir Ahmed, Md Rizwan Parvez, Enamul Hoque, Shafiq Joty Aaryaman Kartha、Ahmed Masry、Mohammed Saidul Islam、Thinh Lang、Shadikur Rahman、Ridwan Mahbub、Mizanur Rahman、Mahir Ahmed、Md Rizwan Parvez、Enamul Hoque、Shafiq JotySubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [279] arXiv:2508.17393 [pdf, html, other]

  Agent-Testing Agent: A Meta-Agent for Automated Testing and Evaluation of Conversational AI Agents Agent-Testing Agent：用于会话式人工智能智能体的自动化测试与评估的元智能体Sameer Komoravolu, Khalil Mrini 萨米尔·科莫拉沃卢，哈利勒·姆里尼Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [280] arXiv:2508.17378 [pdf, html, other]

  UI-Level Evaluation of ALLaM 34B: Measuring an Arabic-Centric LLM via HUMAIN Chat 基于界面的 ALLaM 34B 评估：通过 HUMAIN 聊天衡量一个以阿拉伯语为中心的 LLMOmer NacarSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [281] arXiv:2508.17347 [pdf, html, other]

  The Arabic Generality Score: Another Dimension of Modeling Arabic Dialectness 阿拉伯语通用性评分：阿拉伯方言特性建模的另一个维度Sanad Shaban, Nizar HabashComments: Accepted to EMNLP 2025 Main Conference 注释：已被接收为 EMNLP 2025 主会会议论文Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [282] arXiv:2508.17340 [pdf, html, other]

  Capturing Legal Reasoning Paths from Facts to Law in Court Judgments using Knowledge Graphs 从事实到法律：在判决书中使用知识图谱捕捉法律推理路径Ryoma Kondo, Riona Matsuoka, Takahiro Yoshida, Kazuyuki Yamasawa, Ryohei Hisano 近藤良真、松岡里緒奈、吉田孝洋、山沢一之、久野亮平Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Databases (cs.DB); Information Retrieval (cs.IR) 学科：计算与语言（cs.CL）；人工智能（cs.AI）；数据库（cs.DB）；信息检索（cs.IR）

• [283] arXiv:2508.17337 [pdf, html, other]

  DropLoRA: Sparse Low-Rank Adaptation for Parameter-Efficient Fine-Tuning DropLoRA：用于参数高效微调的稀疏低秩适配Haojie Zhang 张浩杰Comments: 8 pages 注释：8 页Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [284] arXiv:2508.17330 [pdf, other]

  Omne-R1: Learning to Reason with Memory for Multi-hop Question Answering Omne-R1：带记忆的多跳问答推理学习Boyuan Liu, Feng Ji, Jiayan Nan, Han Zhao, Weiling Chen, Shihao Xu, Xing Zhou 刘博远, 季峰, 南佳言, 赵寒, 陈玮玲, 徐世豪, 周行Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [285] arXiv:2508.17324 [pdf, html, other]

  CultranAI at PalmX 2025: Data Augmentation for Cultural Knowledge Representation CultranAI 在 PalmX 2025：用于文化知识表示的数据增强Hunzalah Hassan Bhatti, Youssef Ahmed, Md Arid Hasan, Firoj AlamComments: LLMs, Native, Arabic LLMs, Augmentation, Multilingual, Language Diversity, Contextual Understanding, Minority Languages, Culturally Informed, Foundation Models, Large Language Models 评论：LLMs、本地、阿拉伯语 LLMs、增强、多语言、语言多样性、语境理解、少数语言、文化知情、基础模型、大型语言模型Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [286] arXiv:2508.17310 [pdf, html, other]

  Handling Students Dropouts in an LLM-driven Interactive Online Course Using Language Models 在由 LLM 驱动的交互式在线课程中使用语言模型处理学生辍学问题Yuanchun Wang, Yiyang Fu, Jifan Yu, Daniel Zhang-Li, Zheyuan Zhang, Joy Lim Jia Yin, Yucheng Wang, Peng Zhou, Jing Zhang, Huiqin Liu 王元春，傅一阳，于吉凡，Daniel Zhang-Li，张哲远，Joy Lim Jia Yin，王昱成，周鹏，张静，刘慧琴Comments: 12 pages 备注：12 页Subjects: Computation and Language (cs.CL); Computers and Society (cs.CY) 学科：计算与语言（cs.CL）；计算机与社会（cs.CY）

• [287] arXiv:2508.17281 [pdf, html, other]

  From Language to Action: A Review of Large Language Models as Autonomous Agents and Tool Users 从语言到行动：大语言模型作为自主代理与工具使用者的综述Sadia Sultana Chowa, Riasad Alvi, Subhey Sadi Rahman, Md Abdur Rahman, Mohaimenul Azam Khan Raiaan, Md Rafiqul Islam, Mukhtar Hussain, Sami Azam Sadia Sultana Chowa，Riasad Alvi，Subhey Sadi Rahman，Md Abdur Rahman，Mohaimenul Azam Khan Raiaan，Md Rafiqul Islam，Mukhtar Hussain，Sami AzamComments: 40 pages, 6 figures, 10 tables. Submitted to Artificial Intelligence Review for peer review 备注：40 页，6 幅图，10 张表。已提交给《Artificial Intelligence Review》进行同行评审Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [288] arXiv:2508.17258 [pdf, html, other]

  Are You Sure You’re Positive? Consolidating Chain-of-Thought Agents with Uncertainty Quantification for Aspect-Category Sentiment Analysis 你确定你是积极的吗？通过不确定性量化整合链式思维代理用于方面类别情感分析Filippos Ventirozos, Peter Appleby, Matthew ShardlowComments: 18 pages, 10 figures, 3 tables, Proceedings of the 1st Workshop for Research on Agent Language Models (REALM 2025) 备注：18 页，10 幅图，3 张表，第一届代理语言模型研究研讨会（REALM 2025）论文集Journal-ref: Ventirozos et al. 2025. In Proc. of REALM 2025, pp. 309-326. ACL 期刊引用：Ventirozos 等人，2025 年。发表于 REALM 2025 会议论文集中，第 309-326 页。ACLSubjects: Computation and Language (cs.CL); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；信息检索 (cs.IR)

• [289] arXiv:2508.17250 [pdf, other]

  Routing Distilled Knowledge via Mixture of LoRA Experts for Large Language Model based Bundle Generation 通过 LoRA 专家混合的路由蒸馏知识用于基于大语言模型的捆绑生成Kaidong Feng, Zhu Sun, Hui Fang, Jie Yang, Wenyuan Liu, Yew-Soon Ong 冯开东，孙竹，方辉，杨杰，刘文远，王有顺 (Yew-Soon Ong)Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；信息检索 (cs.IR)

• [290] arXiv:2508.17234 [pdf, html, other]

  ClaimGen-CN: A Large-scale Chinese Dataset for Legal Claim Generation ClaimGen-CN：用于法律主张生成的大规模中文数据集Siying Zhou, Yiquan Wu, Hui Chen, Xavier Hu, Kun Kuang, Adam Jatowt, Ming Hu, Chunyan Zheng, Fei Wu 周思颖，吴奕权，陈晖，胡泽维，匡昆，Adam Jatowt，胡明，郑春燕，吴飞Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [291] arXiv:2508.17225 [pdf, html, other]

  SSFO: Self-Supervised Faithfulness Optimization for Retrieval-Augmented Generation SSFO：面向检索增强生成的自监督可信性优化Xiaqiang Tang, Yi Wang, Keyu Hu, Rui Xu, Chuang Li, Weigao Sun, Jian Li, Sihong Xie 唐夏强，王毅，胡可瑜，徐睿，李闯，孙炜高，李健，谢思宏Comments: Working in progress 备注：工作进行中Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [292] arXiv:2508.17202 [pdf, html, other]

  Active Domain Knowledge Acquisition with $100 Budget: Enhancing LLMs via Cost-Efficient, Expert-Involved Interaction in Sensitive Domains 在具有 100 美元预算的主动领域知识获取：通过成本高效、专家参与的敏感领域交互增强 LLMsYang Wu, Raha Moraffah, Rujing Yao, Jinhong Yu, Zhimin Tao, Xiaozhong Liu 吴扬、Raha Moraffah、姚如静、余金宏、陶志敏、刘晓中Comments: EMNLP 2025 Findings 备注：EMNLP 2025 FindingsSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [293] arXiv:2508.17184 [pdf, html, other]

  Towards Alignment-Centric Paradigm: A Survey of Instruction Tuning in Large Language Models 迈向以对齐为中心的范式：大型语言模型指令微调综述Xudong Han, Junjie Yang, Tianyang Wang, Ziqian Bi, Junfeng Hao, Junhao Song 韩旭东, 杨俊杰, 王天阳, 毕梓倩, 郝俊峰, 宋俊豪Comments: 24 pages, 7 figures, 5 tables 注释：24 页, 7 幅图, 5 张表Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [294] arXiv:2508.17164 [pdf, html, other]

  The Impact of Annotator Personas on LLM Behavior Across the Perspectivism Spectrum 注释者角色对 LLM 行为在观念多元光谱上的影响Olufunke O. Sarumi, Charles Welch, Daniel Braun, Jörg Schlötterer Olufunke O. Sarumi、Charles Welch、Daniel Braun、Jörg SchlöttererComments: Accepted at ICNLSP 2025, Odense, Denmark 备注：已被接收参加 2025 年在丹麦欧登塞举办的 ICNLSP 会议Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [295] arXiv:2508.17162 [pdf, html, other]

  Quantifying Language Disparities in Multilingual Large Language Models 量化多语大型语言模型中的语言差异Songbo Hu, Ivan Vulić, Anna Korhonen 胡松波，伊万·武利奇，安娜·科尔霍宁Comments: Accepted at EMNLP 2025 备注：已被 EMNLP 2025 接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [296] arXiv:2508.17157 [pdf, html, other]

  SPORTSQL: An Interactive System for Real-Time Sports Reasoning and Visualization SPORTSQL：一个用于实时体育推理与可视化的交互式系统Sebastian Martinez, Naman Ahuja, Fenil Bardoliya, Chris Bryan, Vivek Gupta Sebastian Martinez、Naman Ahuja、Fenil Bardoliya、Chris Bryan、Vivek GuptaComments: Under Review at EMNLP 备注：正在 EMNLP 审稿中Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [297] arXiv:2508.17153 [pdf, html, other]

  Natural Language Satisfiability: Exploring the Problem Distribution and Evaluating Transformer-based Language Models 自然语言可满足性：探索问题分布并评估基于 Transformer 的语言模型Tharindu Madusanka, Ian Pratt-Hartmann, Riza Batista-Navarro Tharindu Madusanka，Ian Pratt-Hartmann，Riza Batista-NavarroComments: The paper was accepted to the 62nd Association for Computational Linguistics (ACL 2024), where it won the Best Paper Award 注释：该论文被接收至第 62 届计算语言学协会年会（ACL 2024），并获得最佳论文奖Journal-ref: In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 15278 to 15294. 2024 期刊参考：载于第 62 届计算语言学协会年会论文集（第 1 卷：长文），第 15278 至 15294 页。2024Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [298] arXiv:2508.17148 [pdf, html, other]

  Geolocation-Aware Robust Spoken Language Identification 具有地理定位感知的健壮语音语言识别Qingzheng Wang, Hye-jin Shim, Jiancheng Sun, Shinji WatanabeComments: Accepted to IEEE ASRU 2025. \c{opyright} 2025 IEEE. Personal use permitted. Permission from IEEE required for all other uses including reprinting/republishing, advertising, resale, redistribution, reuse, or creating collective works 评论：已被接收参加 IEEE ASRU 2025。© 2025 IEEE。允许个人使用。对于包括重印/再版、广告、转售、再分发、再利用或创建合集作品在内的所有其他用途，须获得 IEEE 的许可。Subjects: Computation and Language (cs.CL); Sound (cs.SD) 主题：计算与语言（cs.CL）；声音（cs.SD）

• [299] arXiv:2508.17131 [pdf, html, other]

  The Power of Framing: How News Headlines Guide Search Behavior 框架的力量：新闻标题如何引导搜索行为Amrit Poudel, Maria Milkowski, Tim Weninger Amrit Poudel、Maria Milkowski、Tim WeningerComments: Accepted to EMNLP 备注：已被 EMNLP 接受Subjects: Computation and Language (cs.CL); Human-Computer Interaction (cs.HC); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL); 人机交互 (cs.HC); 信息检索 (cs.IR)

• [300] arXiv:2508.17127 [pdf, html, other]

  A Straightforward Pipeline for Targeted Entailment and Contradiction Detection 用于定向蕴含与矛盾检测的直接流程Antonin SulcSubjects: Computation and Language (cs.CL); Logic in Computer Science (cs.LO) 主题：计算与语言（cs.CL）；计算机科学中的逻辑（cs.LO）

• [301] arXiv:2508.17126 [pdf, other]

  Token Homogenization under Positional Bias 令牌在位置偏差下的同质化Viacheslav Yusupov, Danil Maksimov, Ameliia Alaeva, Tatiana Zaitceva, Antipina Anna, Anna Vasileva, Chenlin Liu, Rayuth Chheng, Danil Sazanakov, Andrey Chetvergov, Alina Ermilova, Egor ShvetsovSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [302] arXiv:2508.17078 [pdf, html, other]

  Linguistic Neuron Overlap Patterns to Facilitate Cross-lingual Transfer on Low-resource Languages 促进低资源语言跨语种迁移的语言神经元重叠模式Yuemei Xu, Kexin Xu, Jian Zhou, Ling Hu, Lin Gui 许月梅, 徐科欣, 周剑, 胡玲, 桂林Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [303] arXiv:2508.17057 [pdf, html, other]

  GRAID: Synthetic Data Generation with Geometric Constraints and Multi-Agentic Reflection for Harmful Content Detection GRAID：用于有害内容检测的具有几何约束和多智能体反思的合成数据生成Melissa Kazemi Rad, Alberto Purpura, Himanshu Kumar, Emily Chen, Mohammad Shahed Sorower Melissa Kazemi Rad、Alberto Purpura、Himanshu Kumar、Emily Chen、Mohammad Shahed SorowerComments: 19 pages, 12 figures 评论：19 页，12 幅图Subjects: Computation and Language (cs.CL); Cryptography and Security (cs.CR); Machine Learning (cs.LG) 学科：计算与语言（cs.CL）；密码学与安全（cs.CR）；机器学习（cs.LG）

• [304] arXiv:2508.17028 [pdf, html, other]

  Improving Table Understanding with LLMs and Entity-Oriented Search 使用 LLMs 和面向实体的检索改进表格理解Thi-Nhung Nguyen, Hoang Ngo, Dinh Phung, Thuy-Trang Vu, Dat Quoc NguyenComments: Accepted to COLM 2025 注释：已被 COLM 2025 接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [305] arXiv:2508.17008 [pdf, html, other]

  EduRABSA: An Education Review Dataset for Aspect-based Sentiment Analysis Tasks EduRABSA：用于基于方面的情感分析任务的教育评论数据集Yan Cathy Hua, Paul Denny, Jörg Wicker, Katerina Taskova 颜凯西·华 (Yan Cathy Hua), 保罗·丹尼 (Paul Denny), 约尔格·维克尔 (Jörg Wicker), 卡特琳娜·塔斯科娃 (Katerina Taskova)Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [306] arXiv:2508.17005 [pdf, html, other]

  Planning for Success: Exploring LLM Long-term Planning Capabilities in Table Understanding 为成功而规划：探索 LLM 在表格理解中的长期规划能力Thi-Nhung Nguyen, Hoang Ngo, Dinh Phung, Thuy-Trang Vu, Dat Quoc NguyenComments: Accepted to CoNLL 2025 评注：已被 CoNLL 2025 接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [307] arXiv:2508.17000 [pdf, html, other]

  KL-Regularised Q-Learning: A Token-level Action-Value perspective on Online RLHF KL 正则化的 Q 学习：一种基于令牌级动作值的在线 RLHF 视角Jason R Brown, Lennie Wells, Edward James Young, Sergio Bacallado Jason R Brown、Lennie Wells、Edward James Young、Sergio BacalladoSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [308] arXiv:2508.16998 [pdf, html, other]

  DeAR: Dual-Stage Document Reranking with Reasoning Agents via LLM Distillation DeAR：通过 LLM 蒸馏，以推理代理进行的双阶段文档重排序Abdelrahman Abdallah, Jamshid Mozafari, Bhawna Piryani, Adam Jatowt Abdelrahman Abdallah、Jamshid Mozafari、Bhawna Piryani、Adam JatowtComments: Accept at EMNLP Findings 2025 评审意见：接受于 EMNLP Findings 2025Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；信息检索 (cs.IR)

• [309] arXiv:2508.16994 [pdf, html, other]

  GRADE: Generating multi-hop QA and fine-gRAined Difficulty matrix for RAG Evaluation GRADE：为 RAG 评估生成多跳问答与细粒度难度矩阵Jeongsoo Lee, Daeyong Kwon, Kyohoon Jin 李政洙, 权大勇, 陈庆勋Comments: Accepted at EMNLP 2025 findings 备注：已被 EMNLP 2025 findings 接收Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [310] arXiv:2508.16983 [pdf, html, other]

  ReFactX: Scalable Reasoning with Reliable Facts via Constrained Generation ReFactX：通过受限生成以可靠事实实现可扩展推理Riccardo Pozzi, Matteo Palmonari, Andrea Coletta, Luigi Bellomarini, Jens Lehmann, Sahar Vahdati Riccardo Pozzi、Matteo Palmonari、Andrea Coletta、Luigi Bellomarini、Jens Lehmann、Sahar VahdatiComments: 19 pages, 6 figures, accepted at ISWC 注释：19 页，6 幅图，已被 ISWC 接收Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [311] arXiv:2508.16982 [pdf, html, other]

  Decoding Alignment: A Critical Survey of LLM Development Initiatives through Value-setting and Data-centric Lens 解码对齐：通过价值设定和以数据为中心的视角对 LLM 开发举措的关键综述Ilias Chalkidis 伊利亚斯·查尔基迪斯Comments: This is a working paper and will be updated with new information or corrections based on community feedback 注：这是工作论文，将根据社区反馈更新新信息或更正Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [312] arXiv:2508.16969 [pdf, html, other]

  Explaining Black-box Language Models with Knowledge Probing Systems: A Post-hoc Explanation Perspective 以知识探测系统解释黑箱语言模型：一种事后解释视角Yunxiao Zhao, Hao Xu, Zhiqiang Wang, Xiaoli Li, Jiye Liang, Ru Li 赵云霄，许昊，王志强，李晓莉，梁继业，李茹Comments: 16 pages, 8 figures. This paper has been accepted by DASFAA 2025: The 30th International Conference on Database Systems for Advanced Applications 备注：16 页，8 幅图。该论文已被 DASFAA 2025（第 30 届先进应用数据库系统国际会议）接受Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Databases (cs.DB) 学科：计算与语言（cs.CL）；人工智能（cs.AI）；数据库（cs.DB）

• [313] arXiv:2508.16921 [pdf, other]

  Being Kind Isn’t Always Being Safe: Diagnosing Affective Hallucination in LLMs 善良并不总等同安全：诊断 LLMs 中的情感幻觉Sewon Kim, Jiwon Kim, Seungwoo Shin, Hyejin Chung, Daeun Moon, Yejin Kwon, Hyunsoo YoonComments: 31 pages 备注：31 页Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [314] arXiv:2508.16910 [pdf, html, other]

  Unbiased Reasoning for Knowledge-Intensive Tasks in Large Language Models via Conditional Front-Door Adjustment 通过条件前门调整在大规模语言模型中针对知识密集型任务实现无偏推理Bo Zhao, Yinghao Zhang, Ziqi Xu, Yongli Ren, Xiuzhen Zhang, Renqiang Luo, Zaiwen Feng, Feng XiaComments: This paper has been accepted to the 34th ACM International Conference on Information and Knowledge Management (CIKM 2025), Full Research Paper 注释：本文已被第 34 届 ACM 信息与知识管理国际会议（CIKM 2025）接收，完整研究论文Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [315] arXiv:2508.16889 [pdf, html, other]

  ObjexMT: Objective Extraction and Metacognitive Calibration for LLM-as-a-Judge under Multi-Turn Jailbreaks ObjexMT：用于在多轮越狱情境下以 LLM 作为裁判的目标提取与元认知校准Hyunjun Kim, Junwoo Ha, Sangyoon Yu, Haon ParkSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [316] arXiv:2508.16876 [pdf, html, other]

  Dream to Chat: Model-based Reinforcement Learning on Dialogues with User Belief Modeling Dream to Chat：基于模型的对话强化学习与用户信念建模Yue Zhao, Xiaoyu Wang, Dan Wang, Zhonglin Jiang, Qingqing Gu, Teng Chen, Ningyuan Xi, Jinxian Qu, Yong Chen, Luo Ji 赵岳，王晓雨，王丹，姜中林，顾青青，陈藤，席宁远，曲金仙，陈勇，季洛Comments: Accepted to EMNLP 2025 Findings 评注：已被 EMNLP 2025 Findings 接收Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [317] arXiv:2508.16870 [pdf, html, other]

  JUDGEBERT: Assessing Legal Meaning Preservation Between Sentences JUDGEBERT：评估句子间法律含义的保持情况David Beauchemin, Michelle Albert-Rochette, Richard Khoury, Pierre-Luc Déziel David Beauchemin、Michelle Albert-Rochette、Richard Khoury、Pierre-Luc DézielComments: Accepted to EMNLP 2025 备注：已被 EMNLP 2025 接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [318] arXiv:2508.16867 [pdf, html, other]

  QFrCoLA: a Quebec-French Corpus of Linguistic Acceptability Judgments QFrCoLA：魁北克法语可接受性判断语料库David Beauchemin, Richard KhouryComments: Accepted to EMNLP 2025 备注：已被 EMNLP 2025 接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [319] arXiv:2508.16861 [pdf, html, other]

  Learning from Diverse Reasoning Paths with Routing and Collaboration 从多样化推理路径中学习：路由与协作Zhenyu Lei, Zhen Tan, Song Wang, Yaochen Zhu, Zihan Chen, Yushun Dong, Jundong Li 雷振宇，谭震，王松，朱耀宸，陈子涵，董玉顺，李俊栋Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [320] arXiv:2508.16838 [pdf, html, other]

  If We May De-Presuppose: Robustly Verifying Claims through Presupposition-Free Question Decomposition 如果我们可以去掉预设：通过无预设问题分解对主张进行稳健验证Shubhashis Roy Dipta, Francis Ferraro Shubhashis Roy Dipta，Francis FerraroSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [321] arXiv:2508.16837 [pdf, html, other]

  LLMs Learn Constructions That Humans Do Not Know LLMs 学习人类未知的构式Jonathan Dunn, Mai Mohamed Eida Jonathan Dunn，Mai Mohamed EidaSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [322] arXiv:2508.16833 [pdf, html, other]

  ReProCon: Scalable and Resource-Efficient Few-Shot Biomedical Named Entity Recognition ReProCon：可扩展且资源高效的少样本生物医学命名实体识别Jeongkyun Yoo, Nela Riddle, Andrew Hoblitzell Jeongkyun Yoo、Nela Riddle、Andrew HoblitzellSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [323] arXiv:2508.16788 [pdf, html, other]

  Assess and Prompt: A Generative RL Framework for Improving Engagement in Online Mental Health Communities 评估与提示：一个用于提升在线心理健康社区参与度的生成式强化学习框架Bhagesh Gaur, Karan Gupta, Aseem Srivastava, Manish Gupta, Md Shad Akhtar Bhagesh Gaur、Karan Gupta、Aseem Srivastava、Manish Gupta、Md Shad AkhtarComments: Full Paper accepted in EMNLP Findings 2025 评注：全文被接收于 EMNLP Findings 2025Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [324] arXiv:2508.16762 [pdf, html, other]

  Toward Socially Aware Vision-Language Models: Evaluating Cultural Competence Through Multimodal Story Generation 走向具有社会意识的视觉-语言模型：通过多模态故事生成评估文化能力Arka Mukherjee, Shreya Ghosh Arka Mukherjee，Shreya GhoshComments: Accepted at ASI @ ICCV 2025 备注：已被 ICCV 2025 的 ASI 接收Subjects: Computation and Language (cs.CL); Computers and Society (cs.CY) 学科：计算与语言（cs.CL）；计算机与社会（cs.CY）

• [325] arXiv:2508.16757 [pdf, html, other]

  How Good are LLM-based Rerankers? An Empirical Analysis of State-of-the-Art Reranking Models 基于 LLM 的重排序器有多好？对最先进重排序模型的实证分析Abdelrahman Abdallah, Bhawna Piryani, Jamshid Mozafari, Mohammed Ali, Adam Jatowt Abdelrahman Abdallah、Bhawna Piryani、Jamshid Mozafari、Mohammed Ali、Adam JatowtComments: EMNLP Findings 2025 评论：EMNLP Findings 2025Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；信息检索 (cs.IR)

• [326] arXiv:2508.16753 [pdf, html, other]

  GAICo: A Deployed and Extensible Framework for Evaluating Diverse and Multimodal Generative AI Outputs GAICo：一个已部署且可扩展的框架，用于评估多样化和多模态生成式人工智能输出Nitin Gupta, Pallav Koppisetti, Kausik Lakkaraju, Biplav Srivastava Nitin Gupta，Pallav Koppisetti，Kausik Lakkaraju，Biplav SrivastavaComments: 11 pages, 7 figures, submitted to the Thirty-Eighth Annual Conference on Innovative Applications of Artificial Intelligence (IAAI-26) 注释：11 页，7 幅图，已提交至第三十八届创新人工智能应用年会（IAAI-26）Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [327] arXiv:2508.16729 [pdf, html, other]

  Error Reflection Prompting: Can Large Language Models Successfully Understand Errors? 错误反思提示：大型语言模型能否成功理解错误？Jason Li, Lauren Yraola, Kevin Zhu, Sean O’Brien Jason Li、Lauren Yraola、Kevin Zhu、Sean O’BrienComments: Accepted to Insights @ NAACL 2025 注释：已被接受入选 NAACL 2025 的 InsightsSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [328] arXiv:2508.16707 [pdf, html, other]

  Sparse and Dense Retrievers Learn Better Together: Joint Sparse-Dense Optimization for Text-Image Retrieval 稀疏与密集检索器共同学习能更好：用于文本-图像检索的联合稀疏-密集优化Jonghyun Song, Youngjune Lee, Gyu-Hwung Cho, Ilhyeon Song, Saehun Kim, Yohan Jo Jonghyun Song、Youngjune Lee、Gyu-Hwung Cho、Ilhyeon Song、Saehun Kim、Yohan JoComments: accepted to CIKM 2025 short research paper track 备注：已被 CIKM 2025 短研究论文轨道接收Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR); Machine Learning (cs.LG) 学科：计算与语言（cs.CL）；信息检索（cs.IR）；机器学习（cs.LG）

• [329] arXiv:2508.16705 [pdf, html, other]

  Assessing Consciousness-Related Behaviors in Large Language Models Using the Maze Test 使用迷宫测试评估大语言模型的意识相关行为Rui A. Pimenta, Tim Schlippe, Kristina SchaaffSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [330] arXiv:2508.16697 [pdf, html, other]

  QueryBandits for Hallucination Mitigation: Exploiting Semantic Features for No-Regret Rewriting QueryBandits 用于幻觉缓解：利用语义特征实现无遗憾重写Nicole Cho, William Watson, Alec Koppel, Sumitra Ganesh, Manuela Veloso Nicole Cho，William Watson，Alec Koppel，Sumitra Ganesh，Manuela VelosoSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [331] arXiv:2508.16695 [pdf, html, other] [331] arXiv:2508.16695 [ pdf，html，other ]

  Do Cognitively Interpretable Reasoning Traces Improve LLM Performance? 可认知可解释的推理轨迹是否能提升 LLM 性能？Siddhant Bhambri, Upasana Biswas, Subbarao Kambhampati Siddhant Bhambri、Upasana Biswas、Subbarao KambhampatiSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [332] arXiv:2508.16665 [pdf, html, other]

  Trust but Verify! A Survey on Verification Design for Test-time Scaling 相信但要验证！关于测试时扩展的验证设计综述V Venktesh, Mandeep rathee, Avishek Anand V Venktesh、Mandeep Rathee、Avishek AnandComments: 18 pages 注释：18 页Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [333] arXiv:2508.16636 [pdf, html, other] [333] arXiv:2508.16636 [ pdf，html，other]

  Cognitive Decision Routing in Large Language Models: When to Think Fast, When to Think Slow 大型语言模型中的认知决策路由：何时快速思考，何时慢速思考Y. Du, C. Guo, W. Wang, G. Tang Y. Du，C. Guo，W. Wang，G. TangComments: 6 pages 注释：6 页Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [334] arXiv:2508.16603 [pdf, html, other]

  GreenTEA: Gradient Descent with Topic-modeling and Evolutionary Auto-prompting GreenTEA：结合主题建模与进化自我提示的梯度下降Zheng Dong, Luming Shang, Gabriela OlintoSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [335] arXiv:2508.18234 (cross-list from cs.HC) [pdf, html, other] [335] arXiv:2508.18234（从 cs.HC 交叉列出）[ pdf, html, other]

  Can AI Have a Personality? Prompt Engineering for AI Personality Simulation: A Chatbot Case Study in Gender-Affirming Voice Therapy Training 人工智能能拥有个性吗？为人工智能个性模拟进行提示工程：一个用于性别肯定语音治疗培训的聊天机器人案例研究Tailon D. Jackson, Byunggu Yu Tailon D. Jackson，Byunggu YuSubjects: Human-Computer Interaction (cs.HC); Computation and Language (cs.CL) 主题：人机交互（cs.HC）；计算与语言（cs.CL）

• [336] arXiv:2508.18192 (cross-list from cs.AI) [pdf, html, other] [336] arXiv:2508.18192（跨列表自 cs.AI）[ pdf，html，other]

  Unraveling the cognitive patterns of Large Language Models through module communities 通过模块社区揭示大型语言模型的认知模式Kushal Raj Bhandari, Pin-Yu Chen, Jianxi Gao Kushal Raj Bhandari，Pin-Yu Chen，Jianxi GaoSubjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：人工智能 (cs.AI)；计算与语言 (cs.CL)；机器学习 (cs.LG)

• [337] arXiv:2508.18118 (cross-list from cs.IR) [pdf, html, other] [337] arXiv:2508.18118（从 cs.IR 交叉列表）[ pdf, html, other]

  HLLM-Creator: Hierarchical LLM-based Personalized Creative Generation HLLM-Creator：基于层级 LLM 的个性化创意生成Junyi Chen, Lu Chi, Siliang Xu, Shiwei Ran, Bingyue Peng, Zehuan Yuan 陈钧毅，池璐，徐思亮，冉世伟，彭秉越，袁泽焜Subjects: Information Retrieval (cs.IR); Computation and Language (cs.CL) 学科：信息检索（cs.IR）；计算与语言（cs.CL）

• [338] arXiv:2508.18113 (cross-list from cs.AI) [pdf, html, other] [338] arXiv:2508.18113（跨列自 cs.AI）[ pdf, html, other]

  The AI Data Scientist 人工智能数据科学家Farkhad Akimov, Munachiso Samuel Nwadike, Zangir Iklassov, Martin TakáčSubjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：人工智能 (cs.AI)；计算与语言 (cs.CL)；机器学习 (cs.LG)

• [339] arXiv:2508.18090 (cross-list from cs.DL) [pdf, html, other] [339] arXiv:2508.18090（从 cs.DL 交叉列出）[ pdf, html, other]

  Named Entity Recognition of Historical Text via Large Language Model 通过大型语言模型进行历史文本的命名实体识别Shibingfeng Zhang, Giovanni Colavizza 张世兵峰，Giovanni ColavizzaSubjects: Digital Libraries (cs.DL); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：数字图书馆 (cs.DL)；人工智能 (cs.AI)；计算与语言 (cs.CL)

• [340] arXiv:2508.18006 (cross-list from eess.AS) [pdf, html, other] [340] arXiv:2508.18006（来自 eess.AS 的交叉列表）[ pdf，html，other]

  Unseen Speaker and Language Adaptation for Lightweight Text-To-Speech with Adapters 为轻量级文本到语音系统使用适配器进行未见说话人和语言自适应Alessio Falai, Ziyao Zhang, Akos Gangoly Alessio Falai、Ziyao Zhang、Akos GangolyComments: Accepted at IEEE MLSP 2025 评述：被 IEEE MLSP 2025 接收Subjects: Audio and Speech Processing (eess.AS); Computation and Language (cs.CL); Machine Learning (cs.LG); Sound (cs.SD) 主题：音频与语音处理 (eess.AS)；计算与语言 (cs.CL)；机器学习 (cs.LG)；声音 (cs.SD)

• [341] arXiv:2508.17894 (cross-list from cs.CV) [pdf, html, other] [341] arXiv:2508.17894（来自 cs.CV 的交叉列表）[ pdf, html, other]

  Designing Practical Models for Isolated Word Visual Speech Recognition 为独立词视觉语音识别设计实用模型Iason Ioannis Panagos, Giorgos Sfikas, Christophoros Nikou Iason Ioannis Panagos、Giorgos Sfikas、Christophoros NikouComments: Double-column format, 13 pages with references, 2 figures 注释：双栏格式，13 页含参考文献，2 幅图Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：计算机视觉与模式识别（cs.CV）；人工智能（cs.AI）；计算与语言（cs.CL）

• [342] arXiv:2508.17784 (cross-list from cs.LG) [pdf, html, other] [342] arXiv:2508.17784（来自 cs.LG 的交叉列表）[ pdf，html，other]

  Proximal Supervised Fine-Tuning 近端有监督微调Wenhong Zhu, Ruobing Xie, Rui Wang, Xingwu Sun, Di Wang, Pengfei Liu 朱文宏，谢若冰，王睿，孙兴武，王迪，刘鹏飞Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：机器学习 (cs.LG); 人工智能 (cs.AI); 计算与语言 (cs.CL)

• [343] arXiv:2508.17760 (cross-list from cs.CV) [pdf, html, other] [343] arXiv:2508.17760（来自 cs.CV 的交叉列表）[ pdf，html，other]

  CEIDM: A Controlled Entity and Interaction Diffusion Model for Enhanced Text-to-Image Generation CEIDM：一种用于增强文本到图像生成的受控实体与交互扩散模型Mingyue Yang, Dianxi Shi, Jialu Zhou, Xinyu Wei, Leqian Li, Shaowu Yang, Chunping Qiu 杨明月，史殿西，周佳璐，魏新宇，李乐千，杨少武，邱春平Subjects: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL) 主题：计算机视觉与模式识别 (cs.CV)；计算与语言 (cs.CL)

• [344] arXiv:2508.17753 (cross-list from cs.RO) [pdf, html, other] [344] arXiv:2508.17753（来自 cs.RO 的交叉列表）[ pdf，html，其他]

  Talking to Robots: A Practical Examination of Speech Foundation Models for HRI Applications 与机器人对话：面向人机交互应用的语音基础模型的实用性检验Theresa Pekarek Rosin, Julia Gachot, Henri-Leon Kordt, Matthias Kerzel, Stefan Wermter Theresa Pekarek Rosin、Julia Gachot、Henri-Leon Kordt、Matthias Kerzel、Stefan WermterComments: Accepted at the workshop on Foundation Models for Social Robotics (FoMoSR) at ICSR 2025 备注：已被接收至 2025 年 ICSR 上社会机器人基础模型研讨会（FoMoSR）Subjects: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC) 主题：机器人学 (cs.RO)；人工智能 (cs.AI)；计算与语言 (cs.CL)；人机交互 (cs.HC)

• [345] arXiv:2508.17715 (cross-list from cs.IR) [pdf, html, other] [345] arXiv:2508.17715（从 cs.IR 交叉列出）[ pdf, html, other]

  How Do LLM-Generated Texts Impact Term-Based Retrieval Models? 大型语言模型生成的文本如何影响基于词项的检索模型？Wei Huang, Keping Bi, Yinqiong Cai, Wei Chen, Jiafeng Guo, Xueqi Cheng 黄伟，毕科平，蔡吟穷，陈伟，郭家峰，程学祺Subjects: Information Retrieval (cs.IR); Computation and Language (cs.CL) 学科：信息检索（cs.IR）；计算与语言（cs.CL）

• [346] arXiv:2508.17692 (cross-list from cs.AI) [pdf, html, other] [346] arXiv:2508.17692（从 cs.AI 交叉列出）[ pdf，html，other]

  LLM-based Agentic Reasoning Frameworks: A Survey from Methods to Scenarios 基于 LLM 的主体性推理框架：从方法到场景的综述Bingxi Zhao, Lin Geng Foo, Ping Hu, Christian Theobalt, Hossein Rahmani, Jun Liu 赵炳熙，林更夫，胡平，Christian Theobalt，Hossein Rahmani，刘俊Comments: 51 pages,10 figures,8 tables. Work in progress 备注：51 页，10 幅图，8 张表。进行中工作Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [347] arXiv:2508.17679 (cross-list from cs.LG) [pdf, html, other] [347] arXiv:2508.17679（跨列自 cs.LG）[ pdf, html, other]

  Characterizing the Behavior of Training Mamba-based State Space Models on GPUs 表征在 GPU 上训练基于 Mamba 的状态空间模型的行为Trinayan Baruah, Kaustubh Shivdikar, Sara Prescott, David Kaeli Trinayan Baruah、Kaustubh Shivdikar、Sara Prescott、David KaeliSubjects: Machine Learning (cs.LG); Hardware Architecture (cs.AR); Computation and Language (cs.CL) 主题：机器学习 (cs.LG)；硬件架构 (cs.AR)；计算与语言 (cs.CL)

• [348] arXiv:2508.17638 (cross-list from cs.CV) [pdf, html, other] [348] arXiv:2508.17638（从 cs.CV 交叉列出）[ pdf, html, other]

  Dynamic Embedding of Hierarchical Visual Features for Efficient Vision-Language Fine-Tuning 分层视觉特征的动态嵌入以实现高效的视觉-语言微调Xinyu Wei, Guoli Yang, Jialu Zhou, Mingyue Yang, Leqian Li, Kedi Zhang, Chunping Qiu 魏欣瑜，杨国力，周佳璐，杨明月，李乐谦，张可迪，邱春平Subjects: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL) 主题：计算机视觉与模式识别 (cs.CV)；计算与语言 (cs.CL)

• [349] arXiv:2508.17590 (cross-list from cs.DB) [pdf, html, other] [349] arXiv:2508.17590（从 cs.DB 交叉列出）[ pdf，html，other]

  RubikSQL: Lifelong Learning Agentic Knowledge Base as an Industrial NL2SQL System RubikSQL：作为工业级自然语言到 SQL 系统的终身学习能动知识库Zui Chen, Han Li, Xinhao Zhang, Xiaoyu Chen, Chunyin Dong, Yifeng Wang, Xin Cai, Su Zhang, Ziqi Li, Chi Ding, Jinxu Li, Shuai Wang, Dousheng Zhao, Sanhai Gao, Guangyi Liu 陈最, 李涵, 张鑫浩, 陈晓宇, 董春寅, 王一峰, 蔡鑫, 张肃, 李子奇, 丁驰, 李金旭, 王帅, 赵斗昇, 高三海, 刘光义Comments: 18 pages, 3 figures, 3 tables, to be submitted to VLDB 2026 (PVLDB Volume 19) 注释：18 页，3 幅图，3 张表，拟提交至 VLDB 2026（PVLDB 第 19 卷）Subjects: Databases (cs.DB); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multiagent Systems (cs.MA) 学科：数据库（cs.DB）；人工智能（cs.AI）；计算与语言（cs.CL）；多智能体系统（cs.MA）

• [350] arXiv:2508.17540 (cross-list from cs.LG) [pdf, other] [350] arXiv:2508.17540（从 cs.LG 交叉列表）[ pdf，其他]

  Activation Transport Operators 激活传输算子Andrzej Szablewski, Marek MasiakComments: 4 pages, 4 figures, references and appendices 注释：4 页，4 幅图，参考文献和附录Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：机器学习 (cs.LG); 人工智能 (cs.AI); 计算与语言 (cs.CL)

• [351] arXiv:2508.17445 (cross-list from cs.LG) [pdf, html, other] [351] arXiv:2508.17445（从 cs.LG 交叉列出）[ pdf, html, other]

  TreePO: Bridging the Gap of Policy Optimization and Efficacy and Inference Efficiency with Heuristic Tree-based Modeling TreePO：通过启发式树模型弥合策略优化效果与推理效率之间的差距Yizhi Li, Qingshui Gu, Zhoufutu Wen, Ziniu Li, Tianshun Xing, Shuyue Guo, Tianyu Zheng, Xin Zhou, Xingwei Qu, Wangchunshu Zhou, Zheng Zhang, Wei Shen, Qian Liu, Chenghua Lin, Jian Yang, Ge Zhang, Wenhao HuangSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL) 学科：机器学习（cs.LG）；计算与语言（cs.CL）

• [352] arXiv:2508.17391 (cross-list from cs.AI) [pdf, html, other] [352] arXiv:2508.17391（从 cs.AI 交叉列出）[ pdf，html，other]

  Large Language Models as Universal Predictors? An Empirical Study on Small Tabular Datasets 大型语言模型作为通用预测器？对小型表格数据集的实证研究Nikolaos Pavlidis, Vasilis Perifanis, Symeon Symeonidis, Pavlos S. Efraimidis Nikolaos Pavlidis、Vasilis Perifanis、Symeon Symeonidis、Pavlos S. EfraimidisSubjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [353] arXiv:2508.17334 (cross-list from cs.CV) [pdf, html, other] [353] arXiv:2508.17334（从 cs.CV 交叉列出）[ pdf, html, other]

  Mind the (Language) Gap: Towards Probing Numerical and Cross-Lingual Limits of LVLMs 注意（语言）差距：迈向探查大视觉语言模型的数值与跨语言极限Somraj Gautam, Abhirama Subramanyam Penamakuri, Abhishek Bhandari, Gaurav Harit Somraj Gautam，Abhirama Subramanyam Penamakuri，Abhishek Bhandari，Gaurav HaritSubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算机视觉与模式识别（cs.CV）；人工智能（cs.AI）；计算与语言（cs.CL）；机器学习（cs.LG）

• [354] arXiv:2508.17243 (cross-list from cs.CV) [pdf, html, other] [354] arXiv:2508.17243（从 cs.CV 交叉列出）[ pdf, html, other]

  CoViPAL: Layer-wise Contextualized Visual Token Pruning for Large Vision-Language Models CoViPAL：用于大型视觉-语言模型的逐层上下文化视觉令牌剪枝Zicong Tang, Ziyang Ma, Suqing Wang, Zuchao Li, Lefei Zhang, Hai Zhao, Yun Li, Qianren Wang 唐子聪、马子扬、王素清、李祖超、张乐非、赵海、李云、王千任Comments: Accepted by EMNLP 2025 Findings 备注：被 EMNLP 2025 Findings 接收Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：计算机视觉与模式识别（cs.CV）；人工智能（cs.AI）；计算与语言（cs.CL）

• [355] arXiv:2508.17205 (cross-list from cs.CV) [pdf, html, other] [355] arXiv:2508.17205（来自 cs.CV 的交叉列表）[ pdf, html, other]

  Multi-Agent Visual-Language Reasoning for Comprehensive Highway Scene Understanding 多智能体视觉-语言推理用于综合高速公路场景理解Yunxiang Yang, Ningning Xu, Jidong J. Yang 杨云翔，许宁宁，杨季栋 J.Comments: 16 pages, 16 figures, 8 tables 评论：16 页，16 幅图，8 张表Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Image and Video Processing (eess.IV) 学科：计算机视觉与模式识别（cs.CV）；人工智能（cs.AI）；计算与语言（cs.CL）；图像与视频处理（eess.IV）

• [356] arXiv:2508.17182 (cross-list from cs.LG) [pdf, html, other] [356] arXiv:2508.17182（从 cs.LG 交叉列出）[ pdf, html, other]

  LLM Assertiveness can be Mechanistically Decomposed into Emotional and Logical Components LLM 的自信可以在机制上分解为情感和逻辑成分Hikaru Tsujimura, Arush Tagade 辻村光、Arush TagadeComments: This preprint is under review 注：该预印本正在审稿中Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：机器学习 (cs.LG); 人工智能 (cs.AI); 计算与语言 (cs.CL)

• [357] arXiv:2508.17068 (cross-list from cs.MA) [pdf, other] [357] arXiv:2508.17068（从 cs.MA 交叉列出）[ pdf, 其他]

  Anemoi: A Semi-Centralized Multi-agent System Based on Agent-to-Agent Communication MCP server from Coral Protocol Anemoi：基于代理间通信的半集中式多智能体系统 MCP 服务器，来自 Coral ProtocolXinxing Ren, Caelum Forder, Qianbo Zang, Ahsen Tahir, Roman J. Georgio, Suman Deb, Peter Carroll, Önder Gürcan, Zekun Guo 任新星，Caelum Forder，臧千博，Ahsen Tahir，Roman J. Georgio，Suman Deb，Peter Carroll，Önder Gürcan，郭泽坤Subjects: Multiagent Systems (cs.MA); Computation and Language (cs.CL) 学科：多智能体系统 (cs.MA)；计算与语言 (cs.CL)

• [358] arXiv:2508.17031 (cross-list from cs.SD) [pdf, html, other] [358] arXiv:2508.17031（从 cs.SD 交叉列出）[ pdf, html, other]

  RephraseTTS: Dynamic Length Text based Speech Insertion with Speaker Style Transfer RephraseTTS：基于文本的动态长度语音插入与说话人风格迁移Neeraj Matiyali, Siddharth Srivastava, Gaurav SharmaSubjects: Sound (cs.SD); Computation and Language (cs.CL) 主题：声音 (cs.SD)；计算与语言 (cs.CL)

• [359] arXiv:2508.16936 (cross-list from q-fin.PM) [pdf, html, other] [359] arXiv:2508.16936（从 q-fin.PM 交叉列出）[ pdf, html, other]

  THEME : Enhancing Thematic Investing with Semantic Stock Representations and Temporal Dynamics 主题：通过语义股票表示和时间动态增强主题投资Hoyoung Lee, Wonbin Ahn, Suhwan Park, Jaehoon Lee, Minjae Kim, Sungdong Yoo, Taeyoon Lim, Woohyung Lim, Yongjae Lee 李浩荣、安元彬、朴受焕、李在勋、金旻宰、柳成东、林泰允、林佑炯、李勇宰Comments: Accepted at ACM International Conference on Information and Knowledge Management (CIKM) 备注：已被 ACM 国际信息与知识管理会议（CIKM）接收Subjects: Portfolio Management (q-fin.PM); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Information Retrieval (cs.IR) 主题：投资组合管理 (q-fin.PM)；人工智能 (cs.AI)；计算与语言 (cs.CL)；信息检索 (cs.IR)

• [360] arXiv:2508.16929 (cross-list from cs.LG) [pdf, html, other] [360] arXiv:2508.16929（从 cs.LG 交叉列出）[ pdf, html, other]

  Attention Layers Add Into Low-Dimensional Residual Subspaces 注意力层在低维残差子空间相加Junxuan Wang, Xuyang Ge, Wentao Shu, Zhengfu He, Xipeng Qiu 王君轩，葛旭阳，舒文涛，何正福，裘熙鹏Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL) 学科：机器学习（cs.LG）；计算与语言（cs.CL）

• [361] arXiv:2508.16846 (cross-list from cs.AI) [pdf, html, other] [361] arXiv:2508.16846（跨列表自 cs.AI）[ pdf，html，other]

  Quantifying Sycophancy as Deviations from Bayesian Rationality in LLMs 将谄媚量化为 LLMs 中偏离贝叶斯理性的程度Katherine Atwell, Pedram Heydari, Anthony Sicilia, Malihe Alikhani 凯瑟琳·阿特韦尔，佩德拉姆·海达里，安东尼·西西利亚，马利赫·阿利卡尼Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [362] arXiv:2508.16785 (cross-list from cs.LG) [pdf, html, other] [362] arXiv:2508.16785（从 cs.LG 交叉列出）[ pdf, html, other]

  Interpreting the Effects of Quantization on LLMs 量化对 LLMs 影响的解释Manpreet Singh, Hassan Sajjad 曼普里特·辛格，哈桑·萨贾德Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：机器学习 (cs.LG); 人工智能 (cs.AI); 计算与语言 (cs.CL)

• [363] arXiv:2508.16765 (cross-list from cs.CR) [pdf, html, other] [363] arXiv:2508.16765（来自 cs.CR 的交叉列表）[ pdf, html, other]

  Guarding Your Conversations: Privacy Gatekeepers for Secure Interactions with Cloud-Based AI Models 保护你的对话：面向基于云的人工智能模型的安全交互隐私把关者GodsGift Uzor, Hasan Al-Qudah, Ynes Ineza, Abdul SerwaddaComments: 2025 19th International Conference on Semantic Computing (ICSC) 注释：2025 年第 19 届国际语义计算会议（ICSC）Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：密码学与安全（cs.CR）；人工智能（cs.AI）；计算与语言（cs.CL）

• [364] arXiv:2508.16744 (cross-list from cs.LG) [pdf, html, other] [364] arXiv:2508.16744（来自 cs.LG 的交叉列出）[ pdf, html, other]

  Hyperbolic Multimodal Representation Learning for Biological Taxonomies 用于生物分类法的双曲多模态表示学习ZeMing Gong, Chuanqi Tang, Xiaoliang Huo, Nicholas Pellegrino, Austin T. Wang, Graham W. Taylor, Angel X. Chang, Scott C. Lowe, Joakim Bruslund Haurum 龚泽明，唐传琪，霍晓亮，尼古拉斯·佩莱格里诺，奥斯汀·T·王，格雷厄姆·W·泰勒，安吉尔·X·张，斯科特·C·洛厄，约阿基姆·布鲁斯伦德·豪鲁姆Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV) 学科：机器学习 (cs.LG)；计算与语言 (cs.CL)；计算机视觉与模式识别 (cs.CV)

• [365] arXiv:2508.16681 (cross-list from cs.AI) [pdf, html, other] [365] arXiv:2508.16681（从 cs.AI 交叉列出）[ pdf, html, other]

  Revisiting Rule-Based Stuttering Detection: A Comprehensive Analysis of Interpretable Models for Clinical Applications 重新审视基于规则的口吃检测：面向临床应用的可解释模型综合分析Eric Zhang 埃里克·张Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [366] arXiv:2508.16677 (cross-list from cs.LG) [pdf, html, other] [366] arXiv:2508.16677（从 cs.LG 交叉列出）[ pdf, html, 其他]

  Recall-Extend Dynamics: Enhancing Small Language Models through Controlled Exploration and Refined Offline Integration 回忆-扩展动力学：通过受控探索和精炼的离线整合增强小型语言模型Zhong Guan, Likang Wu, Hongke Zhao, Jiahui Wang, Le WuSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：机器学习 (cs.LG); 人工智能 (cs.AI); 计算与语言 (cs.CL)

• [367] arXiv:2508.16676 (cross-list from cs.LG) [pdf, html, other] [367] arXiv:2508.16676 (从 cs.LG 交叉列出) [ pdf, html, other]

  WISCA: A Lightweight Model Transition Method to Improve LLM Training via Weight Scaling WISCA：一种通过权重缩放改进 LLM 训练的轻量级模型迁移方法Jiacheng Li, Jianchao Tan, Zhidong Yang, Pingwei Sun, Feiye Huo, Jiayu Qin, Yerui Sun, Yuchen Xie, Xunliang Cai, Xiangyu Zhang, Maoxin He, Guangming Tan, Weile Jia, Tong ZhaoSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL) 学科：机器学习（cs.LG）；计算与语言（cs.CL）

• [368] arXiv:2508.16674 (cross-list from cs.CV) [pdf, html, other] [368] arXiv:2508.16674（从 cs.CV 交叉列出）[ pdf, html, other]

  MedRepBench: A Comprehensive Benchmark for Medical Report Interpretation MedRepBench：面向医学报告解读的综合基准Fangxin Shang, Yuan Xia, Dalu Yang, Yahui Wang, Binglin Yang 尚方心，夏元，大禄杨，王雅慧，杨炳霖Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：计算机视觉与模式识别（cs.CV）；人工智能（cs.AI）；计算与语言（cs.CL）

• [369] arXiv:2508.16673 (cross-list from cs.CY) [pdf, html, other] [369] arXiv:2508.16673（从 cs.CY 交叉列出）[ pdf, html, other]

  Invisible Filters: Cultural Bias in Hiring Evaluations Using Large Language Models 隐形过滤器：使用大型语言模型进行招聘评估时的文化偏见Pooja S. B. Rao, Laxminarayen Nagarajan Venkatesan, Mauro Cherubini, Dinesh Babu Jayagopi Pooja S. B. Rao、Laxminarayen Nagarajan Venkatesan、Mauro Cherubini、Dinesh Babu JayagopiComments: Accepted to AIES 2025 评论：已被 AIES 2025 接收Subjects: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：计算机与社会（cs.CY）；人工智能（cs.AI）；计算与语言（cs.CL）

• [370] arXiv:2508.16657 (cross-list from cs.CY) [pdf, other] [370] arXiv:2508.16657（从 cs.CY 跨列）[ pdf，其他]

  Leveraging Multi-Source Textural UGC for Neighbourhood Housing Quality Assessment: A GPT-Enhanced Framework 利用多源文本式用户生成内容进行邻里住房质量评估：一种 GPT 增强框架Qiyuan Hong, Huimin Zhao, Ying Long 洪启源，赵慧敏，龙颖Comments: 6 pages, 3 figures. This paper is reviewed and accepted by the CUPUM (Computational Urban Planning and Urban Management) Conference held by University College London (UCL) in 2025 注释：6 页，3 图。本文已被伦敦大学学院（UCL）于 2025 年主办的 CUPUM（计算城市规划与城市管理）会议审稿通过并接收Subjects: Computers and Society (cs.CY); Computation and Language (cs.CL) 学科：计算机与社会（cs.CY）；计算与语言（cs.CL）

• [371] arXiv:2508.16638 (cross-list from cs.CY) [pdf, html, other] [371] arXiv:2508.16638（来自 cs.CY 的交叉列表）[ pdf, html, other]

  Empirical Analysis of the Effect of Context in the Task of Automated Essay Scoring in Transformer-Based Models 基于 Transformer 模型的自动作文评分任务中上下文影响的实证分析Abhirup ChakravartyComments: MSc Dissertation 评论：硕士论文Subjects: Computers and Society (cs.CY); Computation and Language (cs.CL) 学科：计算机与社会（cs.CY）；计算与语言（cs.CL）

• [372] arXiv:2508.16629 (cross-list from cs.LG) [pdf, html, other] [372] arXiv:2508.16629（从 cs.LG 交叉列出）[ pdf, html, other]

  Learn to Memorize: Optimizing LLM-based Agents with Adaptive Memory Framework 学会记忆：使用自适应记忆框架优化 LLM 基础代理Zeyu Zhang, Quanyu Dai, Rui Li, Xiaohe Bo, Xu Chen, Zhenhua Dong 张泽宇，戴全宇，李睿，薄晓荷，陈旭，董振华Comments: 17 pages, 4 figures, 5 tables 备注：17 页，4 幅图，5 张表Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Information Retrieval (cs.IR) 学科：机器学习 (cs.LG); 人工智能 (cs.AI); 计算与语言 (cs.CL); 信息检索 (cs.IR)

• [373] arXiv:2508.16599 (cross-list from cs.HC) [pdf, other] [373] arXiv:2508.16599（从 cs.HC 交叉列出）[ pdf，其他]

  Humans Perceive Wrong Narratives from AI Reasoning Texts 人类从人工智能推理文本中感知到错误的叙述Mosh Levy, Zohar Elyoseph, Yoav GoldbergSubjects: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人机交互 (cs.HC)；人工智能 (cs.AI)；计算与语言 (cs.CL)

• [374] arXiv:2508.16555 [pdf, html, other]

  Transfer Learning via Lexical Relatedness: A Sarcasm and Hate Speech Case Study 通过词汇相关性进行迁移学习：一个关于讽刺与仇恨言论的案例研究Angelly Cabrera, Linus Lei, Antonio Ortega 安吉莉·卡布雷拉、林纳斯·雷、安东尼奥·奥尔特加Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [375] arXiv:2508.16484 [pdf, html, other]

  HAMSA: Hijacking Aligned Compact Models via Stealthy Automation HAMSA：通过隐蔽自动化劫持对齐紧凑模型Alexey Krylov, Iskander Vagizov, Dmitrii Korzh, Maryam Douiba, Azidine Guezzaz, Vladimir Kokh, Sergey D. Erokhin, Elena V. Tutubalina, Oleg Y. Rogov Alexey Krylov、Iskander Vagizov、Dmitrii Korzh、Maryam Douiba、Azidine Guezzaz、Vladimir Kokh、Sergey D. Erokhin、Elena V. Tutubalina、Oleg Y. RogovComments: 9 pages, 1 figure; article under review 备注：9 页，1 幅图；文章正在审稿中Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [376] arXiv:2508.16478 [pdf, html, other]

  LLM-as-classifier: Semi-Supervised, Iterative Framework for Hierarchical Text Classification using Large Language Models LLM 作为分类器：使用大型语言模型的用于分层文本分类的半监督、迭代框架Doohee You, Andy Parisi, Zach Vander Velden, Lara Dantas Inojosa Doohee You、Andy Parisi、Zach Vander Velden、Lara Dantas InojosaComments: 20 pages excluding reference list, 2 figures 评论：20 页，不包括参考文献列表，2 幅图Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；信息检索 (cs.IR)

• [377] arXiv:2508.16464 [pdf, html, other]

  What makes an entity salient in discourse? 是什么使得在语篇中某个实体显著？Amir Zeldes, Jessica Lin 阿米尔·泽尔德斯, 杰西卡·林Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [378] arXiv:2508.16456 [pdf, html, other]

  A Probabilistic Inference Scaling Theory for LLM Self-Correction 一种用于 LLM 自我纠正的概率推理尺度理论Zhe Yang, Yichang Zhang, Yudong Wang, Ziyao Xu, Junyang Lin, Zhifang Sui 杨哲，张奕昌，王宇栋，徐子尧，林俊阳，隋志方Comments: EMNLP 2025 Main 评论：EMNLP 2025 主会议Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [379] arXiv:2508.16431 [pdf, other]

  Cetvel: A Unified Benchmark for Evaluating Language Understanding, Generation and Cultural Capacity of LLMs for Turkish Cetvel：一个用于评估 LLM 土耳其语理解、生成与文化能力的统一基准Yakup Abrek Er, Ilker Kesen, Gözde Gül Şahin, Aykut ErdemComments: 31 pages, 2 figures, 10 tables 评论：31 页，2 图，10 表Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [380] arXiv:2508.16390 [pdf, html, other]

  MedQARo: A Large-Scale Benchmark for Medical Question Answering in Romanian MedQARo：针对罗马尼亚语医疗问答的大规模基准Ana-Cristina Rogoz, Radu Tudor Ionescu, Alexandra-Valentina Anghel, Ionut-Lucian Antone-Iordache, Simona Coniac, Andreea Iuliana IonescuSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [381] arXiv:2508.16385 [pdf, other]

  ChatGPT-generated texts show authorship traits that identify them as non-human 由 ChatGPT 生成的文本显示出将其识别为非人类的作者特征Vittoria Dentella, Weihang Huang, Silvia Angela Mansi, Jack Grieve, Evelina Leivada Vittoria Dentella、Weihang Huang、Silvia Angela Mansi、Jack Grieve、Evelina LeivadaSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [382] arXiv:2508.16371 [pdf, html, other]

  The Mediomatix Corpus: Parallel Data for Romansh Idioms via Comparable Schoolbooks Mediomatix 语料库：通过可比教科书获取罗曼什语习语的平行数据Zachary Hopton, Jannis Vamvas, Andrin Büchler, Anna Rutkiewicz, Rico Cathomas, Rico Sennrich 扎卡里·霍普顿、雅尼斯·瓦姆瓦斯、安德林·比克勒、安娜·鲁特凯维茨、里科·卡托马斯、里科·森里希Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [383] arXiv:2508.16357 [pdf, html, other]

  MizanQA: Benchmarking Large Language Models on Moroccan Legal Question Answering MizanQA：对摩洛哥法律问答的大型语言模型基准测试Adil Bahaj, Mounir Ghogho 阿迪尔·巴哈杰、穆尼尔·戈霍霍Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；信息检索 (cs.IR)

• [384] arXiv:2508.16325 [pdf, html, other]

  LLMSymGuard: A Symbolic Safety Guardrail Framework Leveraging Interpretable Jailbreak Concepts LLMSymGuard：一个利用可解释越狱概念的符号化安全护栏框架Darpan Aswal, Céline Hudelot Darpan Aswal，Céline HudelotSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Symbolic Computation (cs.SC) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；符号计算（cs.SC）

• [385] arXiv:2508.16303 [pdf, html, other]

  JaParaPat: A Large-Scale Japanese-English Parallel Patent Application Corpus JaParaPat：大规模日英平行专利申请语料库Masaaki Nagata, Katsuki Chousa, Norihito Yasuda 永田正彰、长沙胜树、安田纪仁Comments: LREC-COLING 2024 备注：LREC-COLING 2024Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [386] arXiv:2508.16270 [pdf, html, other]

  LLMs that Understand Processes: Instruction-tuning for Semantics-Aware Process Mining 理解流程的 LLMs：面向语义感知流程挖掘的指令微调Vira Pyrih, Adrian Rebmann, Han van der Aa Vira Pyrih、Adrian Rebmann、Han van der AaComments: Accepted at IEEE ICPM 2025, 8 pages, 2 figures 备注：被 IEEE ICPM 2025 录用，8 页，2 幅图Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [387] arXiv:2508.16267 [pdf, html, other]

  From Confidence to Collapse in LLM Factual Robustness 从置信到崩溃：LLM 事实稳健性的演变Alina Fastowski, Bardh Prenkaj, Gjergji Kasneci Alina Fastowski、Bardh Prenkaj、Gjergji KasneciSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [388] arXiv:2508.16265 [pdf, html, other]

  M3TQA: Massively Multilingual Multitask Table Question Answering M3TQA：大规模多语种多任务表格问答Daixin Shu, Jian Yang, Zhenhe Wu, Xianjie Wu, Xianfu Cheng, Xiangyuan Guan, Yanghai Wang, Pengfei Wu, Tingyang Yang, Hualei Zhu, Wei Zhang, Ge Zhang, Jiaheng Liu, Zhoujun Li Daixin Shu、Jian Yang、Zhenhe Wu、Xianjie Wu、Xianfu Cheng、Xiangyuan Guan、Yanghai Wang、Pengfei Wu、Tingyang Yang、Hualei Zhu、Wei Zhang、Ge Zhang、Jiaheng Liu、Zhoujun LiSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [389] arXiv:2508.16243 [pdf, html, other]

  TULIP: Adapting Open-Source Large Language Models for Underrepresented Languages and Specialized Financial Tasks TULIP：将开源大语言模型适配于代表性不足的语言与专业金融任务İrem Demirtaş, Burak Payzun, Seçil Arslan İrem Demirtaş、Burak Payzun、Seçil ArslanComments: IJCAI 2025 - FinLLM Workshop 注释：IJCAI 2025 - FinLLM 研讨会Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [390] arXiv:2508.16198 [pdf, html, other]

  CMR-SPB: Cross-Modal Multi-Hop Reasoning over Text, Image, and Speech with Path Balance CMR-SPB：基于路径平衡的跨模态多跳推理（文本、图像与语音）Seunghee Kim, Ingyu Bang, Seokgyu Jang, Changhyeon Kim, Sanghwan Bae, Jihun Choi, Richeng Xuan, Taeuk Kim 金昇熙、方仁圭、张锡圭、金昌炫、裴相焕、崔志勋、宣日成、金泰旭Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [391] arXiv:2508.16190 [pdf, html, other]

  ComicScene154: A Scene Dataset for Comic Analysis ComicScene154：用于漫画分析的场景数据集Sandro Paval, Ivan P. Yamshchikov, Pascal MeißnerSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [392] arXiv:2508.16188 [pdf, html, other]

  Seeing is Believing: Emotion-Aware Audio-Visual Language Modeling for Expressive Speech Generation 以眼见为信：情感感知的视听语言建模用于富有表现力的语音生成Weiting Tan, Jiachen Lian, Hirofumi Inaguma, Paden Tomasello, Philipp Koehn, Xutai Ma 谭伟廷、连嘉辰、稻熊浩史、帕登·托马塞洛、菲利普·科恩、马绪泰Comments: EMNLP 2025 (Findings) 备注：EMNLP 2025（Findings）Subjects: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM); Sound (cs.SD); Audio and Speech Processing (eess.AS) 主题：计算与语言 (cs.CL); 计算机视觉与模式识别 (cs.CV); 多媒体 (cs.MM); 声音 (cs.SD); 音频与语音处理 (eess.AS)

• [393] arXiv:2508.16185 [pdf, other]

  ParamBench: A Graduate-Level Benchmark for Evaluating LLM Understanding on Indic Subjects ParamBench：用于评估 LLM 对印度语言学科理解的研究生级基准测试Kaushal Sharma, Vivek Patel, Ayush Maheshwari, Aditya Maheshwari Kaushal Sharma，Vivek Patel，Ayush Maheshwari，Aditya MaheshwariSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [394] arXiv:2508.16139 [pdf, html, other]

  XLQA: A Benchmark for Locale-Aware Multilingual Open-Domain Question Answering XLQA：面向地区感知的多语种开放领域问答基准Keon-Woo Roh, Yeong-Joon Ju, Seong-Whan LeeComments: Accepted to EMNLP 2025 main conference. 12 pages, 4 figures, 7 tables. Code is available at this https URL 注释：已被接受为 EMNLP 2025 主会会议论文。12 页，4 幅图，7 张表。代码可在此 https URL 获取Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [395] arXiv:2508.16122 [pdf, html, other]

  Text Takes Over: A Study of Modality Bias in Multimodal Intent Detection 文本接管：多模态意图检测中的模态偏差研究Ankan Mullick, Saransh Sharma, Abhik Jana, Pawan Goyal Ankan Mullick、Saransh Sharma、Abhik Jana、Pawan GoyalComments: EMNLP 2025 Main Conference Full Paper 备注：EMNLP 2025 主会议全文论文Journal-ref: EMNLP 2025 Main Conference Full Paper 期刊引用：EMNLP 2025 主要会议全文论文Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [396] arXiv:2508.16109 [pdf, html, other]

  From Indirect Object Identification to Syllogisms: Exploring Binary Mechanisms in Transformer Circuits 从间接宾语识别到三段论：在变压器电路中探索二元机制Karim Saraipour, Shichang ZhangSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [397] arXiv:2508.16100 [pdf, html, other]

  CYCLE-INSTRUCT: Fully Seed-Free Instruction Tuning via Dual Self-Training and Cycle Consistency CYCLE-INSTRUCT：通过双重自我训练与循环一致性实现完全无种子指令微调Zhanming Shen, Hao Chen, Yulei Tang, Shaolin Zhu, Wentao Ye, Xiaomeng Hu, Haobo Wang, Gang Chen, Junbo Zhao 沈占明、陈昊、唐玉磊、朱少林、叶文韬、胡晓萌、王浩博、陈刚、赵君博Comments: EMNLP 2025 Main 评论：EMNLP 2025 主会议Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [398] arXiv:2508.16081 [pdf, html, other]

  CEQuest: Benchmarking Large Language Models for Construction Estimation CEQuest：用于施工估算的大型语言模型基准测试Yanzhao Wu, Lufan Wang, Rui Liu 吴言钊，王陆凡，刘睿Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [399] arXiv:2508.16070 [pdf, html, other]

  Less Redundancy: Boosting Practicality of Vision Language Model in Walking Assistants 减少冗余：提升视觉语言模型在步行辅助设备中实用性Chongyang Li, Zhiqiang Yuan, Jiapei Zhang, Ying Deng, Hanbo Bi, Zexi Jia, Xiaoyue Duan, Peixiang Luo, Jinchao Zhang 李重阳、袁志强、张佳培、邓颖、毕汉博、贾泽曦、段晓悦、罗佩翔、张进超Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [400] arXiv:2508.16065 [pdf, html, other]

  Ethical Considerations of Large Language Models in Game Playing 大型语言模型在游戏玩法中的伦理考量Qingquan Zhang, Yuchen Li, Bo Yuan, Julian Togelius, Georgios N. Yannakakis, Jialin Liu 张庆泉、李雨辰、袁博、朱利安·托格里厄斯、乔治奥斯·N·雅纳卡基斯、刘家乐Comments: 19 pages 注释：19 页Journal-ref: Frontiers of Computer Science (2025) 期刊参考：Frontiers of Computer Science (2025)Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [401] arXiv:2508.16048 [pdf, html, other]

  OpenWHO: A Document-Level Parallel Corpus for Health Translation in Low-Resource Languages OpenWHO：面向低资源语言的健康翻译文档级平行语料库Raphaël Merx, Hanna Suominen, Trevor Cohn, Ekaterina Vylomova Raphaël Merx、Hanna Suominen、Trevor Cohn、Ekaterina VylomovaSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [402] arXiv:2508.16021 [pdf, html, other]

  X-Troll: eXplainable Detection of State-Sponsored Information Operations Agents X-Troll：可解释的国家资助信息行动代理检测Lin Tian, Xiuzhen Zhang, Maria Myung-Hee Kim, Jennifer Biggs, Marian-Andrei Rizoiu 林天，张秀珍，Maria Myung-Hee Kim，Jennifer Biggs，Marian-Andrei RizoiuComments: 15 pages, 5 figures, 4 tables, accepted by CIKM2025 注释：15 页，5 张图，4 张表，已被 CIKM2025 接收Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [403] arXiv:2508.16013 [pdf, html, other]

  Political Ideology Shifts in Large Language Models 大语言模型中的政治意识形态转变Pietro Bernardelle, Stefano Civelli, Leon Fröhling, Riccardo Lunardi, Kevin Roitero, Gianluca Demartini Pietro Bernardelle，Stefano Civelli，Leon Fröhling，Riccardo Lunardi，Kevin Roitero，Gianluca DemartiniSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [404] arXiv:2508.15977 [pdf, html, other]

  Dancing with Deer: A Constructional Perspective on MWEs in the Era of LLMs 与鹿共舞：在 LLMs 时代从构式视角看多词表达Claire Bonial, Julia Bonn, Harish Tayyar MadabushiComments: Chapter in Phraseology and Multiword Expressions, Language Science Press (to appear) 注释：收录于《短语学与多词表达》，Language Science Press（即将出版）Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [405] arXiv:2508.15910 [pdf, html, other]

  Evaluating Structured Decoding for Text-to-Table Generation: Evidence from Three Datasets 评估用于文本到表格生成的结构化解码：来自三个数据集的证据Julian Oestreich, Lydia MüllerComments: to be published in the workshop proceedings of the “From Rules to Language Models: Comparative Performance Evaluation” workshop, held alongside RANLP 2025 注：将发表于与 RANLP 2025 同期举办的 “从规则到语言模型：比较性能评估” 研讨会论文集中Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；信息检索 (cs.IR)

• [406] arXiv:2508.15884 [pdf, html, other]

  Jet-Nemotron: Efficient Language Model with Post Neural Architecture Search Jet-Nemotron：通过后期神经架构搜索实现的高效语言模型Yuxian Gu, Qinghao Hu, Shang Yang, Haocheng Xi, Junyu Chen, Song Han, Han Cai 顾宇贤，胡清昊，杨尚，席浩成，陈俊宇，韩松，蔡涵Comments: Tech Report 注：技术报告Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [407] arXiv:2508.15877 [pdf, html, other]

  Annif at the GermEval-2025 LLMs4Subjects Task: Traditional XMTC Augmented by Efficient LLMs Annif 在 GermEval-2025 LLMs4Subjects 任务中的表现：由高效 LLMs 增强的传统 XMTCOsma Suominen, Juho Inkinen, Mona LehtinenComments: 5 pages, 4 figures, accepted at KONVENS 2025. arXiv admin note: substantial text overlap with arXiv:2504.19675 备注：5 页，4 幅图，已被 KONVENS 2025 接收。arXiv 管理员注：与 arXiv:2504.19675 存在大量文字重叠Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Machine Learning (cs.LG) 主题：计算与语言 (cs.CL); 人工智能 (cs.AI); 信息检索 (cs.IR); 机器学习 (cs.LG)

• [408] arXiv:2508.15876 [pdf, html, other]

  DeepMEL: A Multi-Agent Collaboration Framework for Multimodal Entity Linking DeepMEL：一种用于多模态实体链接的多智能体协作框架Fang Wang, Tianwei Yan, Zonghao Yang, Minghao Hu, Jun Zhang, Zhunchen Luo, Xiaoying Bai 王芳，闫天威，杨宗浩，胡明浩，张俊，骆准宸，白晓英Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；多智能体系统 (cs.MA)

• [409] arXiv:2508.15875 [pdf, html, other]

  NEAT: Concept driven Neuron Attribution in LLMs NEAT：在 LLMs 中基于概念的神经元归因Vivek Hruday Kavuri, Gargi Shroff, Rahul Mishra Vivek Hruday Kavuri、Gargi Shroff、Rahul MishraSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [410] arXiv:2508.15868 [pdf, html, other]

  CARFT: Boosting LLM Reasoning via Contrastive Learning with Annotated Chain-of-Thought-based Reinforced Fine-Tuning CARFT：通过带注释的基于思路链的增强微调的对比学习提升 LLM 推理能力Wenqiao Zhu, Ji Liu, Rongjuncheng Zhang, Haipang Wu, Yulun ZhangComments: 14 pages, to appear in EMNLP25 备注：14 页，将发表于 EMNLP25Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [411] arXiv:2508.15861 [pdf, html, other]

  XFinBench: Benchmarking LLMs in Complex Financial Problem Solving and Reasoning XFinBench：在复杂金融问题解决与推理中对 LLMs 的基准测试Zhihan Zhang, Yixin Cao, Lizi Liao 张志涵，曹一鑫，廖立梓Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [412] arXiv:2508.15855 [pdf, html, other]

  Counterspeech for Mitigating the Influence of Media Bias: Comparing Human and LLM-Generated Responses 反言论（Counterspeech）以减轻媒体偏见影响：比较人类与 LLM 生成的回应Luyang Lin, Zijin Feng, Lingzhi Wang, Kam-Fai Wong 林路阳，冯紫瑾，王灵芝，王锦辉Subjects: Computation and Language (cs.CL); Computers and Society (cs.CY); Social and Information Networks (cs.SI) 主题：计算与语言（cs.CL）；计算机与社会（cs.CY）；社会与信息网络（cs.SI）

• [413] arXiv:2508.15854 [pdf, html, other]

  QU-NLP at QIAS 2025 Shared Task: A Two-Phase LLM Fine-Tuning and Retrieval-Augmented Generation Approach for Islamic Inheritance Reasoning QU-NLP 在 QIAS 2025 共享任务：一种用于伊斯兰遗产（继承）推理的两阶段 LLM 微调与检索增强生成方法Mohammad AL-Smadi 穆罕默德·阿尔-斯马迪Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [414] arXiv:2508.15853 [pdf, other]

  MGSC: A Multi-granularity Consistency Framework for Robust End-to-end Asr MGSC：一种用于鲁棒端到端语音识别的多粒度一致性框架Xuwen Yang 杨旭文Comments: 12 pages, 5figures 注释：12 页，5 幅图Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Sound (cs.SD); Audio and Speech Processing (eess.AS) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；声音（cs.SD）；音频与语音处理（eess.AS）

• [415] arXiv:2508.15851 [pdf, html, other]

  DocHop-QA: Towards Multi-Hop Reasoning over Multimodal Document Collections DocHop-QA：面向多模态文档集合的多跳推理Jiwon Park, Seohyun Pyeon, Jinwoo Kim, Rina Carines Cabal, Yihao Ding, Soyeon Caren Han 朴智媛、卞世玹、金晋宇、里娜·卡里内斯·卡巴尔、丁奕昊、韩思妍·卡伦Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [416] arXiv:2508.15849 [pdf, html, other]

  MedCoT-RAG: Causal Chain-of-Thought RAG for Medical Question Answering MedCoT-RAG：用于医学问答的因果思维链检索增强生成（Causal Chain-of-Thought RAG）Ziyu Wang, Elahe Khatibi, Amir M. Rahmani 王子瑜，Elahe Khatibi，Amir M. RahmaniSubjects: Computation and Language (cs.CL); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；信息检索 (cs.IR)

• [417] arXiv:2508.15847 [pdf, html, other]

  Mechanistic Exploration of Backdoored Large Language Model Attention Patterns 对植入后门的大型语言模型注意力模式的机械性探索Mohammed Abu Baker, Lakshmi Babu-Saheer Mohammed Abu Baker，Lakshmi Babu-SaheerComments: 13 pages. Mechanistic analysis of backdoored LLMs (Qwen2.5-3B). Code: this https URL. Base model: unsloth/Qwen2.5-3B-Instruct-unsloth-bnb-4bit. Finetuned models: this https URL 注释：13 页。对带后门的 LLMs（Qwen2.5-3B）进行机械化分析。代码：此 https URL。基础模型：unsloth/Qwen2.5-3B-Instruct-unsloth-bnb-4bit。微调模型：此 https URLSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [418] arXiv:2508.15846 [pdf, html, other]

  CyPortQA: Benchmarking Multimodal Large Language Models for Cyclone Preparedness in Port Operation CyPortQA：为港口运营中的飓风/气旋防备评估多模态大型语言模型的基准Chenchen Kuai, Chenhao Wu, Yang Zhou, Xiubin Bruce Wang, Tianbao Yang, Zhengzhong Tu, Zihao Li, Yunlong Zhang 陈晨块，陈浩武，周洋，王修斌（Bruce Wang），杨天宝，涂正中，李子豪，张云龙Comments: 9 pages, 5 figures 评论：9 页，5 幅图Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [419] arXiv:2508.15845 [pdf, html, other]

  Coarse-to-Fine Personalized LLM Impressions for Streamlined Radiology Reports 面向放射学报告的粗到细个性化 LLM 印象以实现简化Chengbo Sun, Hui Yi Leong, Lei Li 孙承博，梁慧仪，李磊Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [420] arXiv:2508.15842 [pdf, html, other]

  Lexical Hints of Accuracy in LLM Reasoning Chains LLM 推理链中准确性的词汇提示Arne Vanhoyweghen, Brecht Verbeken, Andres Algaba, Vincent GinisComments: 21 pages, 7 figures, 6 tables 注释：21 页，7 图，6 表Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [421] arXiv:2508.15841 [pdf, other]

  A Review of Developmental Interpretability in Large Language Models 大型语言模型发展性可解释性综述Ihor KendiukhovSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [422] arXiv:2508.15837 [pdf, other]

  Statistical Comparative Analysis of Semantic Similarities and Model Transferability Across Datasets for Short Answer Grading 跨数据集短答案评分中语义相似性与模型可迁移性的统计比较分析Sridevi Bonthu, S.Rama Sree, M.H.M. Krishna Prasad Sridevi Bonthu、S.Rama Sree、M.H.M. Krishna PrasadJournal-ref: Int. J. Intell. Syst. Appl. Eng., 12(15s), 530-538, 2024 期刊参考：Int. J. Intell. Syst. Appl. Eng., 12(15s), 530-538, 2024Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [423] arXiv:2508.15836 [pdf, html, other]

  MorphNAS: Differentiable Architecture Search for Morphologically-Aware Multilingual NER MorphNAS：面向形态敏感多语种命名实体识别的可微架构搜索Prathamesh Devadiga, Omkaar Jayadev Shetty, Hiya Nachnani, Prema RSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [424] arXiv:2508.15835 [pdf, other]

  Alvorada-Bench: Can Language Models Solve Brazilian University Entrance Exams? Alvorada-Bench：语言模型能否通过巴西大学入学考试？Henrique GodoySubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [425] arXiv:2508.15834 [pdf, html, other]

  Scalable Scientific Interest Profiling Using Large Language Models 使用大型语言模型进行可扩展的科学兴趣画像Yilun Liang, Gongbo Zhang, Edward Sun, Betina Idnay, Yilu Fang, Fangyi Chen, Casey Ta, Yifan Peng, Chunhua Weng Yilun Liang、Gongbo Zhang、Edward Sun、Betina Idnay、Yilu Fang、Fangyi Chen、Casey Ta、Yifan Peng、Chunhua WengSubjects: Computation and Language (cs.CL); Digital Libraries (cs.DL); Information Retrieval (cs.IR); Other Quantitative Biology (q-bio.OT) 主题：计算与语言 (cs.CL); 数字图书馆 (cs.DL); 信息检索 (cs.IR); 其他定量生物学 (q-bio.OT)

• [426] arXiv:2508.15832 [pdf, html, other]

  A Functionality-Grounded Benchmark for Evaluating Web Agents in E-commerce Domains 用于评估电子商务领域网页代理的基于功能的基准测试Xianren Zhang, Shreyas Prasad, Di Wang, Qiuhai Zeng, Suhang Wang, Wenbo Yan, Mat Hans 张显任, Shreyas Prasad, 王迪, 曾秋海, 王素航, 闫文博, Mat HansComments: 8 pages for main body and 8 pages of appendix 注释：正文 8 页，附录 8 页Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [427] arXiv:2508.15831 [pdf, html, other]

  Who’s Asking? Investigating Bias Through the Lens of Disability Framed Queries in LLMs 谁在提问？通过以残疾为框架的查询在 LLMs 中调查偏见Srikant Panda, Vishnu Hari, Kalpana Panda, Amit Agarwal, Hitesh Laxmichand PatelComments: Preprint 注释：预印本Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；计算机与社会 (cs.CY)

• [428] arXiv:2508.15830 [pdf, html, other]

  DAIQ: Auditing Demographic Attribute Inference from Question in LLMs DAIQ：从问题中审计 LLMs 对人口属性的推断Srikant Panda, Hitesh Laxmichand Patel, Shahad Al-Khalifa, Amit Agarwal, Hend Al-Khalifa, Sharefah Al-GhamdiComments: Preprint 注释：预印本Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [429] arXiv:2508.15829 [pdf, html, other]

  Mining Mental Health Signals: A Comparative Study of Four Machine Learning Methods for Depression Detection from Social Media Posts in Sorani Kurdish 从心理健康信号中挖掘：四种机器学习方法在索拉尼库尔德语社交媒体帖子抑郁检测中的比较研究Idrees Mohammed, Hossein HassaniComments: 13 pages, 4 figures, 5 tables 备注：13 页，4 幅图，5 张表Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [430] arXiv:2508.15827 [pdf, html, other]

  Mini-Omni-Reasoner: Token-Level Thinking-in-Speaking in Large Speech Models Mini-Omni-Reasoner：大型语音模型中的逐标记“言语中思考”Zhifei Xie, Ziyang Ma, Zihang Liu, Kaiyu Pang, Hongyu Li, Jialin Zhang, Yue Liao, Deheng Ye, Chunyan Miao, Shuicheng Yan 谢志飞，马子阳，刘子航，庞凯宇，李宏宇，张嘉霖，廖岳，叶德恒，缪春艳，严水成Comments: Technical report; Work in progress. Project page: this https URL 备注：技术报告；正在进行的工作。项目页面：this https URLSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）；音频与语音处理（eess.AS）

• [431] arXiv:2508.15826 [pdf, other]

  Embarrassed to observe: The effects of directive language in brand conversation 尴尬的观察：品牌对话中指令性语言的影响Andria Andriuzzi, Géraldine Michel Andria Andriuzzi，Géraldine MichelComments: This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made 备注：本开放获取文章遵循知识共享署名-非商业性使用-禁止演绎 许可协议，允许在任何媒介中使用和分发，前提是适当引用原作、用途为非商业性且不进行修改或改编Journal-ref: Psychology & Marketing, Early View (2025) 期刊参考：Psychology & Marketing，Early View（2025）Subjects: Computation and Language (cs.CL); Computers and Society (cs.CY); Human-Computer Interaction (cs.HC); Social and Information Networks (cs.SI) 学科：计算与语言（cs.CL）；计算机与社会（cs.CY）；人机交互（cs.HC）；社交与信息网络（cs.SI）

• [432] arXiv:2508.15825 [pdf, html, other]

  Enhancing Cryptocurrency Sentiment Analysis with Multimodal Features 使用多模态特征提升加密货币情感分析Chenghao Liu, Aniket Mahanti, Ranesh Naha, Guanghao Wang, Erwann Sbai 刘成浩，Aniket Mahanti，Ranesh Naha，王光浩，Erwann SbaiSubjects: Computation and Language (cs.CL); Statistical Finance (q-fin.ST) 学科：计算与语言（cs.CL）；统计金融（q-fin.ST）

• [433] arXiv:2508.15824 [pdf, html, other]

  Avaliação de eficiência na leitura: uma abordagem baseada em PLN 阅读效率评估：一种基于自然语言处理的方法Túlio Sousa de Gois, Raquel Meister Ko. FreitagComments: in Portuguese language, Paper accepted at the XVI Simpósio Brasileiro de Tecnologia da Informação e da Linguagem Humana (STIL 2025) 注释：葡萄牙语，论文被接收于第十六届巴西信息技术与人类语言研讨会（STIL 2025）Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [434] arXiv:2508.15823 [pdf, html, other]

  SDEC: Semantic Deep Embedded Clustering SDEC：语义深度嵌入聚类Mohammad Wali Ur Rahman, Ric Nevarez, Lamia Tasnim Mim, Salim Hariri Mohammad Wali Ur Rahman、Ric Nevarez、Lamia Tasnim Mim、Salim HaririComments: Accepted for publication in IEEE Transactions on Big Data 备注：已被 IEEE Transactions on Big Data 接收发表Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [435] arXiv:2508.15822 [pdf, html, other]

  An Auditable Pipeline for Fuzzy Full-Text Screening in Systematic Reviews: Integrating Contrastive Semantic Highlighting and LLM Judgment 用于系统综述中模糊全文筛查的可审计流程：结合对比语义高亮与 LLM 判断Pouria Mortezaagha, Arya Rahgozar Pouria Mortezaagha，Arya RahgozarSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET); Information Retrieval (cs.IR) 学科：计算与语言（cs.CL）；人工智能（cs.AI）；新兴技术（cs.ET）；信息检索（cs.IR）

• [436] arXiv:2508.15820 [pdf, other]

  Research on intelligent generation of structural demolition suggestions based on multi-model collaboration 基于多模型协作的结构拆除建议智能生成研究Zhifeng Yang, Peizong Wu 杨志峰，吴培宗Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；信息检索 (cs.IR)

• [437] arXiv:2508.15817 [pdf, html, other]

  Meet Your New Client: Writing Reports for AI – Benchmarking Information Loss in Market Research Deliverables 认识你的新客户：为人工智能撰写报告——市场研究交付物信息损失基准测试Paul F. Simmering, Benedikt Schulz, Oliver Tabino, Georg Wittenburg Paul F. Simmering，Benedikt Schulz，Oliver Tabino，Georg WittenburgComments: 16 pages, 4 figures, 3 tables 注：16 页，4 幅图，3 张表Subjects: Computation and Language (cs.CL); Computers and Society (cs.CY) 学科：计算与语言（cs.CL）；计算机与社会（cs.CY）

• [438] arXiv:2508.15815 [pdf, html, other]

  User-Assistant Bias in LLMs LLMs 中的用户-助手偏差Xu Pan, Jingxuan Fan, Zidi Xiong, Ely Hahami, Jorin Overwiening, Ziqian Xie 徐盼，范静轩，熊子弟，埃利·哈哈米，乔林·欧维宁，谢子谦Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC) 主题：计算与语言 (cs.CL)；人工智能 (cs.AI)；人机交互 (cs.HC)

• [439] arXiv:2508.15813 [pdf, html, other]

  SCOPE: A Generative Approach for LLM Prompt Compression SCOPE：用于 LLM 提示压缩的生成式方法Tinghui Zhang, Yifan Wang, Daisy Zhe Wang 张廷辉，王一凡，王哲（Daisy Zhe Wang）Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [440] arXiv:2508.15811 [pdf, html, other]

  From Clicks to Preference: A Multi-stage Alignment Framework for Generative Query Suggestion in Conversational System 从点击到偏好：用于对话系统中生成式查询建议的多阶段对齐框架Junhao Yin, Haolin Wang, Peng Bao, Ju Xu, Yongliang Wang 尹俊豪，王浩霖，宝鹏，徐菊，王永良Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [441] arXiv:2508.15810 [pdf, html, other]

  Detecting Hope, Hate, and Emotion in Arabic Textual Speech and Multi-modal Memes Using Large Language Models 使用大型语言模型检测阿拉伯语文本语音和多模态表情包中的希望、仇恨与情感Nouar AlDahoul, Yasir Zaki Nouar AlDahoul，Yasir ZakiComments: 26 pages, 12 figures 注释：26 页，12 幅图Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [442] arXiv:2508.15809 [pdf, html, other] [442] arXiv:2508.15809 [ pdf，html，其他]

  Chain-of-Query: Unleashing the Power of LLMs in SQL-Aided Table Understanding via Multi-Agent Collaboration Chain-of-Query：通过多代理协作在 SQL 辅助的表格理解中释放 LLMs 的力量Songyuan Sui, Hongyi Liu, Serena Liu, Li Li, Soo-Hyun Choi, Rui Chen, Xia Hu 宋源遂、刘宏毅、刘赛琳、李莉、崔洙贤、陈睿、胡霞Comments: 9 pages main content, 24 pages total including appendix, 6 figures 注释：正文 9 页，包含附录共 24 页，6 幅图Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Databases (cs.DB) 学科：计算与语言（cs.CL）；人工智能（cs.AI）；数据库（cs.DB）

• [443] arXiv:2508.15807 [pdf, html, other]

  KL-based self-distillation for large language models 基于 KL 的用于大语言模型的自蒸馏Max Rehman LinderComments: Master’s thesis 注：硕士论文Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [444] arXiv:2508.15806 [pdf, html, other]

  SurfaceLogicKV: Surface and Logic Attention Behaviors are All You Need for Robust KV Cache Compression SurfaceLogicKV：表面和逻辑注意力行为是实现鲁棒键值缓存压缩的全部所需Mengjie Li, William J. Song 李梦洁，威廉·J·宋Comments: 18 pages, 9 tables, 10 pages 注释：18 页，9 张表格，10 页Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [445] arXiv:2508.15805 [pdf, html, other]

  ALAS: Autonomous Learning Agent for Self-Updating Language Models ALAS：用于自我更新语言模型的自主学习代理Dhruv AtrejaSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [446] arXiv:2508.15804 [pdf, html, other]

  ReportBench: Evaluating Deep Research Agents via Academic Survey Tasks ReportBench：通过学术调查任务评估深度研究代理Minghao Li, Ying Zeng, Zhihao Cheng, Cong Ma, Kai JiaSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [447] arXiv:2508.15802 [pdf, html, other]

  MAC: A Live Benchmark for Multimodal Large Language Models in Scientific Understanding MAC：用于科学理解的多模态大语言模型实时基准测试Mohan Jiang, Jin Gao, Jiahao Zhan, Dequan WangSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [448] arXiv:2508.15801 [pdf, other]

  LingVarBench: Benchmarking LLM for Automated Named Entity Recognition in Structured Synthetic Spoken Transcriptions LingVarBench：用于结构化合成语音转录中自动命名实体识别的 LLM 基准测试Seyedali Mohammadi, Manas Paldhe, Amit Chhabra Seyedali Mohammadi、Manas Paldhe、Amit ChhabraComments: 10 pages 备注：10 页Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG) 学科：计算与语言（cs.CL）；人工智能（cs.AI）；人机交互（cs.HC）；机器学习（cs.LG）

• [449] arXiv:2508.15800 [pdf, html, other]

  A BERT-based Hierarchical Classification Model with Applications in Chinese Commodity Classification 基于 BERT 的层次分类模型及其在中文商品分类中的应用Kun Liu, Tuozhen Liu, Feifei Wang, Rui PanComments: 29 pages, 3 figures, and 8 tables 注释：29 页，3 幅图，8 张表Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [450] arXiv:2508.15799 [pdf, html, other]

  A Framework for Processing Textual Descriptions of Business Processes using a Constrained Language – Technical Report 用于处理业务流程文本描述的受限语言框架 —— 技术报告Andrea Burattin, Antonio Grama, Ana-Maria Sima, Andrey Rivkin, Barbara Weber Andrea Burattin、Antonio Grama、Ana-Maria Sima、Andrey Rivkin、Barbara WeberSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [451] arXiv:2508.15798 [pdf, html, other]

  Persuasiveness and Bias in LLM: Investigating the Impact of Persuasiveness and Reinforcement of Bias in Language Models LLM 的说服力与偏见：研究说服力与偏见强化对语言模型的影响Saumya RoySubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [452] arXiv:2508.15797 [pdf, html, other]

  Benchmarking the Medical Understanding and Reasoning of Large Language Models in Arabic Healthcare Tasks 在阿拉伯语医疗任务中对大型语言模型的医学理解与推理进行基准测试Nouar AlDahoul, Yasir ZakiComments: 5 pages, 2 figures 评论：5 页，2 幅图Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；人工智能（cs.AI）；机器学习（cs.LG）

• [453] arXiv:2508.15796 [pdf, html, other]

  Benchmarking the Legal Reasoning of LLMs in Arabic Islamic Inheritance Cases 在阿拉伯语伊斯兰继承案例中对 LLMs 法律推理能力的基准测试Nouar AlDahoul, Yasir ZakiComments: 5 pages, 3 figures 注释：5 页，3 幅图Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Machine Learning (cs.LG) 学科：计算与语言（cs.CL）；人工智能（cs.AI）；计算机与社会（cs.CY）；机器学习（cs.LG）

• [454] arXiv:2508.15794 [pdf, html, other]

  Do Language Models Agree with Human Perceptions of Suspense in Stories? 语言模型与人类对故事悬念感知是否一致？Glenn Matlin, Devin Zhang, Rodrigo Barroso Loza, Diana M. Popescu, Joni Isbell, Chandreyi Chakraborty, Mark Riedl Glenn Matlin、Devin Zhang、Rodrigo Barroso Loza、Diana M. Popescu、Joni Isbell、Chandreyi Chakraborty、Mark RiedlJournal-ref: Published at the Conference on Language Models (COLM) 2025 期刊参考：发表于语言模型会议（COLM）2025Subjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [455] arXiv:2508.15793 [pdf, html, other]

  Format as a Prior: Quantifying and Analyzing Bias in LLMs for Heterogeneous Data 作为先验的格式：量化与分析 LLMs 在异构数据上的偏见Jiacheng Liu, Mayi Xu, Qiankun Pi, Wenli Li, Ming Zhong, Yuanyuan Zhu, Mengchi Liu, Tieyun Qian 刘佳成、徐马依、皮千坤、李文利、钟鸣、朱媛媛、刘孟驰、钱铁云Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG) 主题：计算与语言（cs.CL）；机器学习（cs.LG）

• [456] arXiv:2508.15792 [pdf, html, other]

  Bhav-Net: Knowledge Transfer for Cross-Lingual Antonym vs Synonym Distinction via Dual-Space Graph Transformers Bhav-Net：通过双空间图变换器进行跨语言反义词与同义词区分的知识迁移Samyak S. SanghviSubjects: Computation and Language (cs.CL) 学科：计算与语言（cs.CL）

• [457] arXiv:2508.15791 [pdf, html, other]

  InteChar: A Unified Oracle Bone Character List for Ancient Chinese Language Modeling InteChar：用于古代中文语言建模的统一甲骨文字表Xiaolei Diao, Zhihan Zhou, Lida Shi, Ting Wang, Ruihua Qi, Hao Xu, Daqian Shi 肖磊 刁，周志寒，石丽达，王霆，齐瑞华，徐浩，史大谦Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [458] arXiv:2508.15790 [pdf, html, other]

  KG-o1: Enhancing Multi-hop Question Answering in Large Language Models via Knowledge Graph Integration KG-o1：通过知识图谱集成增强大模型的多跳问答能力Nan Wang, Yongqi Fan, yansha zhu, ZongYu Wang, Xuezhi Cao, Xinyan He, Haiyun Jiang, Tong Ruan, Jingping Liu 王楠，樊永琪，朱彦莎，王宗宇，曹学志，何欣妍，江海云，阮彤，刘景平Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

• [459] arXiv:2508.16560 (cross-list from cs.LG) [pdf, html, other] [459] arXiv:2508.16560（跨列表自 cs.LG）[ pdf, html, other]

  Sparse but Wrong: Incorrect L0 Leads to Incorrect Features in Sparse AutoencodersDavid Chanin, Adrià Garriga-AlonsoSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：机器学习 (cs.LG); 人工智能 (cs.AI); 计算与语言 (cs.CL)

• [460] arXiv:2508.16514 (cross-list from cs.LG) [pdf, html, other] [460] arXiv:2508.16514（跨列表自 cs.LG）[ pdf, html, other]

  FLAMES: Improving LLM Math Reasoning via a Fine-Grained Analysis of the Data Synthesis Pipeline FLAMES：通过对数据合成管道的细粒度分析提升 LLM 的数学推理能力Parker Seegmiller, Kartik Mehta, Soumya Saha, Chenyang Tao, Shereen Oraby, Arpit Gupta, Tagyoung Chung, Mohit Bansal, Nanyun Peng Parker Seegmiller、Kartik Mehta、Soumya Saha、Chenyang Tao、Shereen Oraby、Arpit Gupta、Tagyoung Chung、Mohit Bansal、Nanyun PengComments: To appear at EMNLP 2025 注释：将发表于 EMNLP 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：机器学习 (cs.LG); 人工智能 (cs.AI); 计算与语言 (cs.CL)

• [461] arXiv:2508.16453 (cross-list from cs.SI) [pdf, html, other] [461] arXiv:2508.16453（从 cs.SI 跨栏）[ pdf，html，other]

  Anti-establishment sentiment on TikTok: Implications for understanding influence(rs) and expertise on social media TikTok 上的反体制情绪：对理解社交媒体上影响力者和专业性的意义Tianliang Xu, Ariel Hasell, Sabina Tomkins 许天亮, Ariel Hasell, Sabina TomkinsComments: 10 pages excluding references; 14 pages in total; 4 figures; Accepted by the AAAI Conference on Web and Social Media (ICWSM-2026) 评论：不含参考文献 10 页；总计 14 页；4 幅图；被 AAAI 网络与社会媒体会议（ICWSM-2026）接受Subjects: Social and Information Networks (cs.SI); Computation and Language (cs.CL); Machine Learning (cs.LG) 学科：社会与信息网络（cs.SI）；计算与语言（cs.CL）；机器学习（cs.LG）

• [462] arXiv:2508.16439 (cross-list from cs.CY) [pdf, html, other] [462] arXiv:2508.16439（从 cs.CY 交叉列出）[ pdf, html, other]

  PediatricsMQA: a Multi-modal Pediatrics Question Answering Benchmark PediatricsMQA：一个多模态儿科问答基准Adil Bahaj, Oumaima Fadi, Mohamed Chetouani, Mounir Ghogho Adil Bahaj、Oumaima Fadi、Mohamed Chetouani、Mounir GhoghoSubjects: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Graphics (cs.GR); Multimedia (cs.MM) 学科：计算机与社会（cs.CY）；人工智能（cs.AI）；计算与语言（cs.CL）；图形学（cs.GR）；多媒体（cs.MM）

• [463] arXiv:2508.16406 (cross-list from cs.CR) [pdf, other] [463] arXiv:2508.16406（从 cs.CR 交叉列出）[ pdf, other]

  Retrieval-Augmented Defense: Adaptive and Controllable Jailbreak Prevention for Large Language Models 检索增强的防御：针对大型语言模型的自适应与可控越狱防护Guangyu Yang, Jinghong Chen, Jingbiao Mei, Weizhe Lin, Bill Byrne 杨广宇，陈靖鸿，梅靖彪，林维哲，Bill ByrneSubjects: Cryptography and Security (cs.CR); Computation and Language (cs.CL) 主题：密码学与安全（cs.CR）；计算与语言（cs.CL）

• [464] arXiv:2508.16402 (cross-list from cs.SE) [pdf, html, other] [464] arXiv:2508.16402（跨列表自 cs.SE）[ pdf，html，other]

  AetherCode: Evaluating LLMs’ Ability to Win In Premier Programming Competitions AetherCode：评估 LLMs 在顶级编程比赛中取胜的能力Zihan Wang, Jiaze Chen, Zhicheng Liu, Markus Mak, Yidi Du, Geonsik Moon, Luoqi Xu, Aaron Tua, Kunshuo Peng, Jiayi Lu, Mingfei Xia, Boqian Zou, Chenyang Ran, Guang Tian, Shoutai Zhu, Yeheng Duan, Zhenghui Kang, Zhenxing Lin, Shangshu Li, Qiang Luo, Qingshen Long, Zhiyong Chen, Yihan Xiao, Yurong Wu, Daoguang Zan, Yuyi Fu, Mingxuan Wang, Ming Ding 王梓涵、陈家泽、刘志成、Markus Mak、杜奕迪、文建锡、徐洛奇、Aaron Tua、彭坤硕、陆佳怡、夏明飞、邹博千、冉晨阳、田光、朱守泰、段业衡、康郑辉、林振兴、李尚书、罗强、龙庆申、陈志勇、肖一涵、吴雨蓉、昝道广、富雨逸、王明轩、丁鸣Comments: 15 pages 注释：15 页Subjects: Software Engineering (cs.SE); Computation and Language (cs.CL) 学科：软件工程（cs.SE）；计算与语言（cs.CL）

• [465] arXiv:2508.16332 (cross-list from cs.SD) [pdf, html, other] [465] arXiv:2508.16332（从 cs.SD 交叉列出）[ pdf, html, other]

  Vevo2: Bridging Controllable Speech and Singing Voice Generation via Unified Prosody Learning Vevo2：通过统一韵律学习连接可控语音与人声演唱生成Xueyao Zhang, Junan Zhang, Yuancheng Wang, Chaoren Wang, Yuanzhe Chen, Dongya Jia, Zhuo Chen, Zhizheng WuComments: We will release code and model checkpoints at this https URL 评论：我们将在此 https URL 发布代码和模型检查点Subjects: Sound (cs.SD); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：声学 (cs.SD)；人工智能 (cs.AI)；计算与语言 (cs.CL)

• [466] arXiv:2508.16313 (cross-list from cs.LG) [pdf, html, other] [466] arXiv:2508.16313（从 cs.LG 交叉列出）[ pdf, html, other]

  Retrieval Enhanced Feedback via In-context Neural Error-book 通过上下文的神经错误簿增强检索反馈Jongyeop Hyun, Bumsoo Kim Jongyeop Hyun、Bumsoo KimComments: Accepted at EMNLP 2025 main conference 评论：被接收为 EMNLP 2025 主会议论文Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：机器学习 (cs.LG); 人工智能 (cs.AI); 计算与语言 (cs.CL)

• [467] arXiv:2508.16201 (cross-list from cs.CV) [pdf, html, other] [467] arXiv:2508.16201（来自 cs.CV 的交叉分类）[ pdf，html，其他]

  SpecVLM: Enhancing Speculative Decoding of Video LLMs via Verifier-Guided Token Pruning SpecVLM：通过验证器引导的令牌剪枝提升视频 LLMs 的猜测式解码Yicheng Ji, Jun Zhang, Heming Xia, Jinpeng Chen, Lidan Shou, Gang Chen, Huan Li 季一成，张骏，夏鹤鸣，陈锦鹏，寿立丹，陈刚，李欢Comments: Accepted at EMNLP 2025 Main 备注：已被接收于 EMNLP 2025 主会Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 学科：计算机视觉与模式识别（cs.CV）；人工智能（cs.AI）；计算与语言（cs.CL）

• [468] arXiv:2508.16153 (cross-list from cs.LG) [pdf, html, other] [468] arXiv:2508.16153（从 cs.LG 交叉列出）[ pdf, html, other]

  Memento: Fine-tuning LLM Agents without Fine-tuning LLMs Memento：在不微调 LLMs 的情况下微调 LLM 代理Huichi Zhou, Yihang Chen, Siyuan Guo, Xue Yan, Kin Hei Lee, Zihan Wang, Ka Yiu Lee, Guchun Zhang, Kun Shao, Linyi Yang, Jun Wang 周慧驰，陈逸航，郭思远，闫雪，李建熙，王子涵，李嘉耀，张古春，邵坤，杨林毅，王俊Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL) 学科：机器学习（cs.LG）；计算与语言（cs.CL）

• [469] arXiv:2508.16151 (cross-list from cs.AR) [pdf, html, other] [469] arXiv:2508.16151（从 cs.AR 交叉列出）[ pdf, html, other]

  Hardwired-Neurons Language Processing Units as General-Purpose Cognitive Substrates 硬连线神经元语言处理单元作为通用认知基底Yang Liu, Yi Chen, Yongwei Zhao, Yifan Hao, Zifu Zheng, Weihao Kong, Zhangmai Li, Dongchen Jiang, Ruiyang Xia, Zhihong Ma, Zisheng Liu, Zhaoyong Wan, Yunqi Lu, Ximing Liu, Hongrui Guo, Zhihao Yang, Zhe Wang, Tianrui Ma, Mo Zou, Rui Zhang, Ling Li, Xing Hu, Zidong Du, Zhiwei Xu, Qi Guo, Tianshi Chen, Yunji Chen 杨柳，陈毅，赵永伟，郝一帆，郑子甫，孔伟豪，李章迈，姜东辰，夏瑞阳，马志宏，刘紫升，万朝勇，陆云祺，刘希明，郭宏睿，杨志豪，王喆，马天睿，邹墨，张瑞，李玲，胡兴，杜子东，徐志伟，郭琦，陈天时，陈云霁Subjects: Hardware Architecture (cs.AR); Computation and Language (cs.CL) 学科：硬件架构 (cs.AR)；计算与语言 (cs.CL)

• [470] arXiv:2508.16117 (cross-list from cs.AI) [pdf, html, other] [470] arXiv:2508.16117（来自 cs.AI 的交叉列表）[ pdf, html, other]

  Extending FKG.in: Towards a Food Claim Traceability Network 扩展 FKG.in：走向食品声明可追溯网络Saransh Kumar Gupta, Rizwan Gulzar Mir, Lipika Dey, Partha Pratim Das, Anirban Sen, Ramesh Jain Saransh Kumar Gupta、Rizwan Gulzar Mir、Lipika Dey、Partha Pratim Das、Anirban Sen、Ramesh JainComments: 10 pages, 3 figures, 1 table, 45 references, ACM International Conference on Multimedia 2025 - Multi-modal Food Computing Workshop 注释：10 页，3 图，1 表，45 篇参考文献，ACM 国际多媒体会议 2025 - 多模态食品计算研讨会Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Information Retrieval (cs.IR) 主题：人工智能（cs.AI）；计算与语言（cs.CL）；信息检索（cs.IR）

• [471] arXiv:2508.16054 (cross-list from cs.AI) [pdf, other] [471] arXiv:2508.16054（跨列自 cs.AI）[ pdf，其他]

  Generative Foundation Model for Structured and Unstructured Electronic Health Records 用于结构化和非结构化电子健康记录的生成式基础模型Sonish Sivarajkumar, Hang Zhang, Yuelyu Ji, Maneesh Bilalpur, Xizhi Wu, Chenyu Li, Min Gu Kwak, Shyam Visweswaran, Yanshan Wang Sonish Sivarajkumar、Hang Zhang、Yuelyu Ji、Maneesh Bilalpur、Xizhi Wu、Chenyu Li、Min Gu Kwak、Shyam Visweswaran、Yanshan WangSubjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：人工智能（cs.AI）；计算与语言（cs.CL）

• [472] arXiv:2508.15940 (cross-list from cs.AR) [pdf, other] [472] arXiv:2508.15940（跨列自 cs.AR）[ pdf，其他]

  ASIC-Agent: An Autonomous Multi-Agent System for ASIC Design with Benchmark Evaluation ASIC-Agent：用于 ASIC 设计的自主多智能体系统及基准评估Ahmed Allam, Youssef Mansour, Mohamed ShalanComments: 2025 IEEE International Conference on LLM-Aided Design (ICLAD) 注释：2025 年 IEEE 国际 LLM 辅助设计会议（ICLAD）Subjects: Hardware Architecture (cs.AR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Distributed, Parallel, and Cluster Computing (cs.DC); Multiagent Systems (cs.MA) 主题：硬件架构（cs.AR）；人工智能（cs.AI）；计算与语言（cs.CL）；分布式、并行与集群计算（cs.DC）；多智能体系统（cs.MA）

• [473] arXiv:2508.15882 (cross-list from cs.SD) [pdf, html, other] [473] arXiv:2508.15882（从 cs.SD 交叉列出）[ pdf, html, other]

  Beyond Transcription: Mechanistic Interpretability in ASR 超越转录：自动语音识别中的机械可解释性Neta Glazer, Yael Segal-Feldman, Hilit Segev, Aviv Shamsian, Asaf Buchnick, Gill Hetz, Ethan Fetaya, Joseph Keshet, Aviv NavonSubjects: Sound (cs.SD); Computation and Language (cs.CL); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS) 主题：声音 (cs.SD)；计算与语言 (cs.CL)；机器学习 (cs.LG)；音频与语音处理 (eess.AS)

• [474] arXiv:2508.15878 (cross-list from cs.LO) [pdf, html, other] [474] arXiv:2508.15878（从 cs.LO 交叉列出）[ pdf, html, other]

  Lean Meets Theoretical Computer Science: Scalable Synthesis of Theorem Proving Challenges in Formal-Informal Pairs Lean 遇见理论计算机科学：在形式-非形式对中可扩展的定理证明挑战合成Terry Jingchen Zhang, Wenyuan Jiang, Rongchuan Liu, Yisong Wang, Junran Yang, Ning Wang, Nicole Ni, Yinya Huang, Mrinmaya Sachan Terry Jingchen Zhang、Wenyuan Jiang、Rongchuan Liu、Yisong Wang、Junran Yang、Ning Wang、Nicole Ni、Yinya Huang、Mrinmaya SachanComments: Accepted to AI4MATH@ICML2025 注释：已被 AI4MATH@ICML2025 接收Subjects: Logic in Computer Science (cs.LO); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG) 学科：计算机科学中的逻辑 (cs.LO)；人工智能 (cs.AI)；计算与语言 (cs.CL)；机器学习 (cs.LG)

• [475] arXiv:2508.15859 (cross-list from q-bio.NC) [pdf, html, other] [475] arXiv:2508.15859（从 q-bio.NC 交叉列出）[ pdf，html，other]

  Beyond Individuals: Collective Predictive Coding for Memory, Attention, and the Emergence of Language 超越个体：用于记忆、注意力和语言起源的集体预测编码Tadahiro Taniguchi 谷口忠大Journal-ref: Cognitive Neuroscience, 1-2 (2025) 期刊引用：Cognitive Neuroscience，1-2（2025）Subjects: Neurons and Cognition (q-bio.NC); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) 主题：神经元与认知 (q-bio.NC)；人工智能 (cs.AI)；计算与语言 (cs.CL)

• [476] arXiv:2508.15852 (cross-list from cs.LG) [pdf, other] [476] arXiv:2508.15852（从 cs.LG 交叉列出）[ pdf，其他]

  PGF-Net: A Progressive Gated-Fusion Framework for Efficient Multimodal Sentiment Analysis PGF-Net：一种用于高效多模态情感分析的渐进门控融合框架Bin Wen, Tien-Ping Tan 温斌，陈天平Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL) 学科：机器学习（cs.LG）；计算与语言（cs.CL）

• [477] arXiv:2508.15848 (cross-list from cs.CR) [pdf, html, other] [477] arXiv:2508.15848（从 cs.CR 交叉列出）[ pdf, html, other]

  Self-Disguise Attack: Induce the LLM to disguise itself for AIGT detection evasion 自我伪装攻击：诱导 LLM 伪装自身以规避 AIGT 检测Yinghan Zhou, Juan Wen, Wanli Peng, Zhengxian Wu, Ziwei Zhang, Yiming Xue 周英涵，温娟，彭万里，吴正贤，张子炜，薛一鸣Subjects: Cryptography and Security (cs.CR); Computation and Language (cs.CL) 主题：密码学与安全（cs.CR）；计算与语言（cs.CL）

• [478] arXiv:2508.15840 (cross-list from cs.CR) [pdf, html, other] [478] arXiv:2508.15840（从 cs.CR 交叉列出）[ pdf, html, other]

  Unveiling Unicode’s Unseen Underpinnings in Undermining Authorship Attribution 揭示 Unicode 在破坏作者归属中的隐秘基础Robert Dilworth 罗伯特·迪尔沃思Subjects: Cryptography and Security (cs.CR); Computation and Language (cs.CL); Information Retrieval (cs.IR) 学科：密码学与安全（cs.CR）；计算与语言（cs.CL）；信息检索（cs.IR）

• [479] arXiv:2508.15828 (cross-list from cs.LG) [pdf, html, other] [479] arXiv:2508.15828（从 cs.LG 交叉列出）[ pdf, html, 其他]

  Z-Pruner: Post-Training Pruning of Large Language Models for Efficiency without Retraining Z-Pruner：大规模语言模型的后训练剪枝，以在无需再训练的情况下提高效率Samiul Basir Bhuiyan, Md. Sazzad Hossain Adib, Mohammed Aman Bhuiyan, Muhammad Rafsan Kabir, Moshiur Farazi, Shafin Rahman, Nabeel Mohammed Samiul Basir Bhuiyan、Md. Sazzad Hossain Adib、Mohammed Aman Bhuiyan、Muhammad Rafsan Kabir、Moshiur Farazi、Shafin Rahman、Nabeel MohammedComments: Accepted at AICCSA 2025 备注：已被 AICCSA 2025 接收Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL) 学科：机器学习（cs.LG）；计算与语言（cs.CL）

• [480] arXiv:2504.11695 (cross-list from cs.CV) [pdf, html, other] [480] arXiv:2504.11695（跨列表自 cs.CV）[ pdf，html，other]

  Interpreting the linear structure of vision-language model embedding spaces 解读视觉-语言模型嵌入空间的线性结构Isabel Papadimitriou, Huangyuan Su, Thomas Fel, Sham Kakade, Stephanie Gil Isabel Papadimitriou、Huangyuan Su、Thomas Fel、Sham Kakade、Stephanie GilComments: COLM 2025 备注：COLM 2025Subjects: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL); Multimedia (cs.MM) 主题：计算机视觉与模式识别（cs.CV）；计算与语言（cs.CL）；多媒体（cs.MM）

1.2.2 Artificial Intelligence

From：https://papers.cool/arxiv/cs.AI

From：https://arxiv.org/list/cs.AI/recenthttps://arxiv.org/list/cs.CL/recent

1.3 Huggingface

• InternVL3.5：在通用性、推理和效率方面推进开源多模态模型（114▲）
• Visual-CoG：基于引导链的阶段感知强化学习用于文本到图像的生成（34▲）
• MV-RAG：检索增强多视图扩散（25▲）
• twi - reasonbench：基准推理-知情文本到图像生成（20▲）
• MMTok: vlm高效推理的多模态覆盖最大化（18▲）
• 突破探索瓶颈：用于一般LLM推理的规则框架强化学习（17▲）
• 超越记忆：用递归、记忆和测试时间计算缩放扩展推理深度（15▲）
• PosterGen：通过多代理llm生成具有审美意识的纸张到海报（9▲）
• UQ：基于未解决问题的语言模型评估（7▲）
• MEENA（波斯语mmmu）： n级多模态多语言教育考试（6▲）
• 爱马仕4技术报告（5▲）
• 先解释后回答：构图视觉推理综述（2▲）
• 还有10多篇论文
• 基于推理分解的自奖励视觉语言模型（61▲）
• 超越转录：ASR的机制可解释性（49▲）
• 离散扩散VLA：将离散扩散引入视觉-语言-动作策略的动作解码（19▲）
• MIDAS：通过实时自回归视频生成的多模态交互式数字人合成（18▲）
• CODA：用解耦强化学习协调双脑计算机使用代理的大脑和小脑（13▲）
• 预测即将到来的标记的顺序可以改进语言建模（13▲）
• StepWiser：智能推理的逐步生成法官（10▲）
• 扩散语言模型在解码之前就知道答案（10▲）
• 小心第三只眼！ 基于mlm的智能手机代理的隐私意识基准测试（9▲）
• AudioStory：用大语言模型生成长篇叙事音频（7▲）
• MotionFlux：通过整流匹配和偏好对齐高效文本引导运动生成（6▲）
• 凝视心脏：用于rPPG和健康生物标志物估计的多视图视频数据集（5▲）
• 还有5篇论文
• TreePO：用启发式树建模弥合政策优化与有效性和推理效率的差距（57▲）
• CMPhysBench：用于评估凝聚态物理中大语言模型的基准（42▲）
• VibeVoice技术报告（37▲）
• VoxHammer：无需训练的精确和连贯的3D编辑在本地3D空间（28▲）
• OmniHuman-1.5：通过认知模拟在化身中灌输积极的思想（26▲）
• UltraMemV2：内存网络扩展到120B参数与优越的长上下文学习（21▲）
• Pixie：基于像素的3D物理快速、可推广的监督学习（17▲）
• CineScale：高分辨率电影视觉生成的免费午餐（14▲）
• Wan-S2V：音频驱动的电影视频生成（10▲）
• 自回归通用视频分割模型（10▲）
• 基于推理增强语言模型的可解释药物批准预测（9▲）
• 垫片：走向工程科学启示（9▲）
• 还有14篇论文
• InternVL3.5：在通用性、推理和效率方面推进开源多模态模型（114▲）
• Visual-CoG：基于引导链的阶段感知强化学习用于文本到图像的生成（34▲）
• MV-RAG：检索增强多视图扩散（25▲）
• twi - reasonbench：基准推理-知情文本到图像生成（20▲）
• MMTok: vlm高效推理的多模态覆盖最大化（18▲）
• 突破探索瓶颈：用于一般LLM推理的规则框架强化学习（17▲）
• 超越记忆：用递归、记忆和测试时间计算缩放扩展推理深度（15▲）
• PosterGen：通过多代理llm生成具有审美意识的纸张到海报（9▲）
• UQ：基于未解决问题的语言模型评估（7▲）
• MEENA（波斯语mmmu）： n级多模态多语言教育考试（6▲）
• 爱马仕4技术报告（5▲）
• 先解释后回答：构图视觉推理综述（2▲）
• 还有10多篇论文
• AgentFly：没有微调LLM的微调LLM代理（54▲）
• 奥德赛：开放世界四足动物探索和操作的长期视界任务（34▲）
• Beyond Pass@1：用变分问题综合的自我游戏维持RLVR（27▲）
• EgoTwin：第一人称的梦想身体和视角（15▲）
• AetherCode：评估LLMs在顶级编程竞赛中的获胜能力（9▲）
• TPLA：高效分解预填充&解码推理的张量并行潜在注意（5▲）
• 面向可追溯诊断推理的端到端代理RAG系统培训（5▲）
• 做什么? 教授视觉-语言-行动模型，拒绝不可能（5▲）
• AgentScope 1.0：构建代理应用程序的以开发者为中心的框架（5▲）
• 弱监督可视性基础的选择性对比学习（4▲）
• 蒸馏- 3dgs：蒸馏3D高斯飞溅（4▲）
• CRISP：基于稀疏自编码器的持续概念学习（3▲）
• 还有6篇论文
• 实习- s1：科学的多模态基础模型（165▲）
• 自信地深入思考（25▲）
• LiveMCP-101：对具有挑战性的查询进行压力测试和诊断启用mcp的代理（25▲）
• Mobile-Agent-v3: GUI自动化的基础代理（24▲）
• Wave: Wave Your Way to life - like Video Generation（13▲）
• SceneGen：单图像3D场景生成在一个前馈通道（11▲）
• 大型语言模型基准调查（7▲）
• 基于视觉自回归建模的指导图像编辑（6▲）
• ATLAS：用于表达参数化人体建模的骨骼和形状参数解耦（5▲）
• aiXiv：人工智能科学家生成的新一代科学发现开放获取生态系统（4▲）
• “咖啡馆的入口看起来方便吗？” 门在哪儿？” 面向视觉查询的地理空间AI代理（3▲）
• Snap-Snap：用两张图像在毫秒内重建三维人体高斯分布（2▲）
• 还有4篇论文
• DuPO：通过双偏好优化实现可靠的LLM自我验证（54▲）
• 从分数到技能：评估金融大型语言模型的认知诊断框架（51▲）
• FutureX: LLM代理在未来预测中的高级实时基准（45▲）
• MeshCoder：从点云生成llm驱动的结构化网格代码（27▲）
• Tinker：扩散给3D的礼物——从稀疏输入到无场景优化的多视图一致编辑（24▲）
• 从AI for Science到代理科学：自主科学发现述评（19▲）
• MCP-Universe：使用真实世界模型上下文协议服务器对大型语言模型进行基准测试（16▲）
• 量化与dLLMs相遇：扩散LLMs训练后量化的系统研究（14▲）
• NVIDIA Nemotron Nano 2：精确高效的混合曼巴-变压器推理模型（13▲）
• RynnEC：将mlm带入实体世界（9▲）
• FLARE：快速低级别注意力路由引擎（3▲）
• 视觉语言模型在越南多模态考试问题上比人类更好吗？ (3)▲
• 还有6篇论文
• 代理链：基于多代理蒸馏和代理RL的端到端代理基础模型（65▲）
• LongSplat：用于休闲长视频的鲁棒无定位3D高斯溅射（41▲）
• 提示业务流程标记语言（23▲）
• MultiRef：具有多个视觉参考的可控图像生成（15▲）
• 使用配置文件感知的llm作为法官评估播客推荐（10▲）
• emboded - r1：通用机器人操作的强化具身推理（9▲）
• OmniTry：虚拟试戴任何没有面具的东西（9▲）
• 注意生成过程：LLM生成过程中的细粒度置信估计（8▲）
• 使用多模态扩散变压器进行无训练文本引导的颜色编辑（8▲）
• 语音分离的进展：技术、挑战和未来趋势(8)
• 利用大型语言模型对人类苦难进行预测分析（8▲）
• 一针省九针：语言模型的主动自我改进（7▲）
• 还有16篇论文
• Ovis2.5技术报告（75▲）
• ComoRAG：一个认知启发的记忆组织的有状态长篇叙事推理RAG（46▲）
• 4DNeX：前馈4D生成建模变得容易（43▲）
• 下一代视觉粒度生成（38▲）
• 速度总是赢家：大型语言模型的高效架构调查（33▲）
• 当标点符号很重要：LLMs提示鲁棒性方法的大规模比较（32▲）
• GPT-5实现了空间智能吗？ 实证研究（23▲）
• herbench：虚拟世界中长期规划和结构化推理的基准（21）
• 矩阵游戏2.0：一个开源、实时、流媒体的互动世界模型（15▲）
• 利用人工耳蜗标记的自回归预测来表示语音（12▲）
• 流明：视频生成模型中一致的视频重照明和和谐的背景替换（9▲）
• G-CUT3R：引导三维重建与相机和深度先验集成（9▲）
• 还有6篇论文

1.4 X

1.5 小红书

1. 逐步式：逐步生成式判别机制实现更明智推理
2. 思维链CoT再遭质疑！三大证据实锤，真正可泛化推理还很远？

2. 感兴趣研究

正则表达式删除无关字符串

\[PDF\d*\] \[Copy\d*\] \[Kimi\d*\] \[REL\d*\]

\[PDF( \d+)? \] \[复制\] \[Kimi( \d+)? \] \[(?:REL|相关)\]

图片插入

![](https://gitee.com/dujh22/pic/raw/master/logicReason/SLR.png)

人机协作

破解人机协作密码：工作技能拆成两层，AI执行人类决策成功率狂飙 | ICML 2025

游戏

GPT-5通关《宝可梦水晶》创纪录！9517步击败赤爷，效率碾压o3三倍！

刚刚，大模型棋王诞生！40轮血战，OpenAI o3豪夺第一，人类大师地位不保？

矩阵游戏2.0：一个开源、实时、流媒体的互动世界模型（15▲）

模型

刚刚，马斯克开源Grok 2.5：中国公司才是xAI最大对手

字节突然开源Seed-OSS，512K上下文碾压主流4倍长度！推理能力刷新纪录

实测DeepSeek V3.1，不止拓展上下文长度

Gemini 3 本周发布？

合成数据

合成数据的「毒」与「药」，模型崩溃有何新解？

评测

We-Math 2.0：全新多模态数学推理数据集 × 首个综合数学知识体系

大型语言模型基准调查（7▲）

数学

比GPT-5还准？AIME25飙到99.9%刷屏，开源模型首次！

突破探索瓶颈：用于一般LLM推理的规则框架强化学习（17▲）

超越记忆：用递归、记忆和测试时间计算缩放扩展推理深度（15▲）

自进化

GPT-5点赞！八大顶尖机构发布「自进化智能体」全面综述

一针省九针：语言模型的主动自我改进（7▲）

学习

Hugging Face 推出九大 AI 课程，免费、全面【收藏】

推理泛化性

思维链CoT再遭质疑！三大证据实锤，真正可泛化推理还很远？

自我博弈

🧬SvS-自我博弈问题合成，助力RL突破极限

智能体

代理链：基于多代理蒸馏和代理RL的端到端代理基础模型（65▲）

逻辑推理

[82] arXiv:2508.19903 [pdf, html, other]

Logical Reasoning with Outcome Reward Models for Test-Time Scaling 用于测试时扩展的带有结果奖励模型的逻辑推理Ramya Keerthy Thatikonda, Wray Buntine, Ehsan ShareghiComments: EMNLP 2025 评论：EMNLP 2025Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) 学科：计算与语言（cs.CL）；人工智能（cs.AI）

[474] arXiv:2508.15878 (cross-list from cs.LO) [pdf, html, other] [474] arXiv:2508.15878（从 cs.LO 交叉列出）[ pdf, html, other]

Lean Meets Theoretical Computer Science: Scalable Synthesis of Theorem Proving Challenges in Formal-Informal Pairs Lean 遇见理论计算机科学：在形式-非形式对中可扩展的定理证明挑战合成Terry Jingchen Zhang, Wenyuan Jiang, Rongchuan Liu, Yisong Wang, Junran Yang, Ning Wang, Nicole Ni, Yinya Huang, Mrinmaya Sachan Terry Jingchen Zhang、Wenyuan Jiang、Rongchuan Liu、Yisong Wang、Junran Yang、Ning Wang、Nicole Ni、Yinya Huang、Mrinmaya SachanComments: Accepted to AI4MATH@ICML2025 注释：已被 AI4MATH@ICML2025 接收Subjects: Logic in Computer Science (cs.LO); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG) 学科：计算机科学中的逻辑 (cs.LO)；人工智能 (cs.AI)；计算与语言 (cs.CL)；机器学习 (cs.LG)